I am creating a wiki for the answer(s), please feel free to update below each question as appropriate. If you have a second answer, number it accordingly and place it below other answers.
For (a) an application that has NOT yet been issued, sub-questions:
(1) Does the NEW application cancel the previous (original) application, taking its priority (filing) date? If the new application is defeated/rejected, does their original application come back to life or is it also lost?
A continuation is a later application filed during the pendency of an earlier application and which has the same specification (disclosure) as the prior application. Usually this is done to change (improve) the claims. The continuation does not necessarily mean the original application is dead, although most often the applicant, having improved the claims, no longer needs the prior application and abandons it. Normally that abandonment is expressly done in a transmittal letter accompanying the continuation when filed. If the continuation adds new subject matter to the specification, it is called a "continuation in part" since only part is the same specification. If the continuation has claims to a different invention, it is called a "divisional" rather than a continuation, since a divisional carves out or divides out a different invention found in the same specification. Just like any other application, either the first application (called the "parent" case) or the continuation, or both, may be searched, examined, rejected, or allowed by the Patent Office if the basic, search and examination fees were paid, and either the parent case, the continuation case, or both, may be abandoned or issue as a patent. So it's possible that both the original application and the continuation would result in (separate) patents being issued.
Think of the original application as a lit candle that stays lit as long as the application is pending. So long as the candle is lit, it can be used to light other candles. Those other candles are the continuation, divisional or continuation-in-part applications. Once the two applications are being processed, they are each judged on their merits. US patent law allows applicants to file as many continuations as they like and in whatever series. The only restriction is that you need a lit candle (pending application) to light another candle. And each candle can be considered to have been lit on the date the first candle was lit.
(2) Does a significant change in the continuation application mean that the original application(s) was/were flawed and/or incomplete? How does this affect the new patent application credibility?
No, it doesn't mean the original was necessarily flawed or incomplete. Rather, there are many varied strategies used by patent applicants when building a patent family. Most often continuations result from improvements to the invention or a desire to add claims after examination has been concluded. This might happen where the applicant discovers a product that might be covered by the claims but wants to be sure and so files additional claims in a continuation to specifically cover that product. That product might be the applicant's product or might be a competitive product that the applicant wants to stop as an infringement. As noted in (1) above, the continuation may be what is called a divisional to cover a separate invention disclosed in the parent case. There is some times strategic sense to including multiple inventions in a single case and then filing divisional cases. In the lit candle example, this might be where the first candle was described as red, white or blue and the examiner says those are 3 separate inventions, so the applicant "elects" to prosecute the red candle in the first case and files 2 divisional applications, one to a white candle and one to a blue candle. Three candle patents might result from this.
(3) Assuming the continuation application builds upon the prior application (ergo, requiring prior art pertaining to original claims to predate the previous filing date), how does prior art on ADDED or CHANGED items affect the application, i.e. does prior art for new/changed claims also need to predate the ORIGINAL filing date?
As to all matter disclosed in the parent case, a continuation gets the priority date of the parent application from which it is derived. An applicant can have a continuation of a continuation of a continuation in part of a ..., going on for years) and each will get an "effective filing date" of the first filed application in the chain. Just for one example, US Patent 6,442,336 says:
"This application is a continuation of application Ser. No. 08/110,861, filed Aug. 24, 1993, now U.S. Pat. No. 5,446,599 which is a division of application Ser. No. 07/818,168, filed Jan. 8, 1992, now U.S. Pat. No. 5,260,837, which is a division of application Ser. No. 07/325,768, filed Mar. 20, 1989, abandoned, which is a continuation-in-part of application Ser. No. 06/877,319, filed Jun. 23, 1986, now U.S. Pat. No. 4,819,101, which is a continuation of application Ser. No. 06/516,532, filed Jul. 25, 1983, now U.S. Pat. No. 4,604,668, which is a continuation of application Ser. No. 06/209,057, filed Nov. 21, 1980, abandoned."
So, that patent would have an effective filing date of Nov. 21, 1980. However, it is important to note that, since 1995, the term of a patent is 20 years from its effective filing date, while prior to that date the term was 17 years from the issue date. The change was made to eliminate long pendencies and suprises when such "submarine patents" surfaced after decades to potentially sink later products. This was famously the case with Jerome Lemelson and his "machine vision" patents which were pending more than 20 years and subsequently issued and resulted in more than a billion dollars of royalty for Lemelson from manufacturers of bar code scanners before the patents were eventually invalidated for abuse of the patent system due to the long pendency.
For (b) an issued patent that HAS expired:
(1) Can a continuation patent be filed for an expired patent?
To be a continuation, an application must be filed prior to issuance of the application from which it seeks to claim priority. In the case of a chain of continuations, so long as one application in the chain is still pending, it is possible to file a continuation, even if a grandparent or great-great-great grandparent case has already issued, and even if that first patent has expired. However, this is not practical because the term of a patent is 20 years from the effective filing date, so claiming the benefit of the filing date of an expired patent means claiming back more than 20 years, and that would mean the continuation case so claiming would have zero term, it would expire immediately upon issuance and thus provide no exclusive rights. So, yes you could do it, but you would be a fool if you did.
For (c) an issued patent that has NOT expired, sub-questions:
(1) Can a trivial change be used to simply extend the life of the patent, indefinitely by applying trivial changes periodically? Is there a solid litmus test for this or is it just up to the examiners?
Yes and no. You can continue an application, but only while that application remains in prosecution (i.e., before it issues as a patent). Under the old patent expiration model (i.e., that patents were valid for 17 years after being issued) yes, it was possible to extend the life of a patent for quite a while by writing continuations, continuations-in-part, divisionals, etc. Under the current system (that a patent expires 20 years after application) this has been drastically reduced, if not eliminated entirely, because (at least normally) the date of the original application will be used to compute the expiration date of a continuation [see prior answer for "expired patents".].
(2) Can a major change that effectively turns the invention into something else be introduced periodically, resulting in indefinite continuations by simply writing in or incorporating new technologies as they come out, thus broadening the scope of the original invention (to the point where it covers completely different industries/areas)?
Yes. That is what continuation-in-part applications ("CIPs") seek to accomplish. However, once a patent expires, the subject matter of its claims becomes public domain, so only the new aspects introduced in the CIPs still unexpired would thereafter be possible to cover or protect through CIPs. People have certainly tried to "stretch" an application to cover newer technologies. Just for example, Jerome Lemelson (the inventor on the patent quoted above) once wrote a patent on a "computer system". After TI came out with the first microcontroller (i.e., essentially a complete computer system on a single integrated circuit), Mr. Lemelson wrote a continuation of one of his earlier patents that added something to the effect that: "...and all of it is on one integrated circuit". The patent issued, but was eventually invalidated -- the court found that the original specification contained no support for the idea of the whole system being on a single integrated circuit, so the TI team (Gary Boone, and, if memory serves, a few others) officially invented the microcontroller.
(3) If a continuation patent is found to be invalid, how does that affect the original (built upon) patent(s)?
They're independent patents -- each is valid or invalid on its own. That said, if any patent in the chain of continuations is found invalid due to insufficient teaching of the patented invention or due to prior art, there's probably an increased chance that the same would be true of any derived patents as well.