@Toxaris - FlightCrew has come up previously in the thread but I've never given it any serious thought.
Here are my opinions - and of course you or others may disagree, as I said I have used neither tool enough to know what one does that the other doesn't. However personally I find such tools to have close to zero value, for the simple reason that they produce too many false positives and noise. Every ePub pretty much will have some form of "Error" on nearly every html page. However for 99.99% of users out there these "errors" will have zero impact on the readability of the document and are of no interest whatsoever. That is not to say that there aren't checks in QC that don't visibly affect readability either - however *you* decide which things are important to you and are reported.
For instance I just ran ePubCheck against a retail ePub, and got a whole screen of errors. Every html page in the epub has a <meta name="Adept.resource value="xxx"/> tag in it. And ePubCheck flags this as an "error" because according to it value is not an allowed attribute there. If every ePub in your library is being flagged with these sort of nonsense "errors", to me it is a tool that adds zero value to a workflow. As a user you have to start filtering the output to visually ignore things. Then users would start saying "can't QC give me some sort of GUI to control which things get reported as errors" and basically it would end up becoming a beast in its own right with workarounds to compensate for a poorly designed tool. Add to that it is java and a complication to deploy/run and for me ePubCheck is a complete non-starter.
FlightCrew of course is now part of Sigil and I see its last release was Aug 2011. I have no idea what the roadmap for FlightCrew is - has it become a dead-end product now that it is in Sigil, and only Sigil will get any updates to it? Or is it continuing to be maintained but just hasn't had a need for a release?
From my quick experiments it seems to produce even more "noise" than ePubCheck - giving two error message lines for each occurrence of that value meta tag above. So again you would be battling the fact that every single ePub is going to show as "errored" and deal with tens of thousands of lines of log file output if you ran it across your library. On the plus side for FlightCrew vs ePubCheck it is better packaged/easier to run, and has the good sense to have error numbers in its output. So a gui for this tool to control the output would be more practical - assuming you were willing to trawl the source code to find all the error numbers and figure out whether they were things the average user would or would not care about. You would arguably be better off having a "FlightCrew" plugin for calibre with its own gui for controlling and working with it?
I also found it to be extremely slow to execute - at a rough guess about 7 seconds for a single ePub. So if you wanted to run it against a number of books in your library it would be come a "leave it overnight" activity. And if you are just running it one book at a time, why not hit Alt+E with the Open With plugin to bring it up in Sigil and validate it there? At least that way you can edit the Epub there and then to fix errors that you care about and revalidate it. Otherwise you have to trawl a log file, search for the book, and then launch Sigil...
I'm not saying "never" on FlightCrew for QC, I just need to be convinced it could be incorporated in a way that would add value. I'm sure there are genuine important errors that such tools identify, they just get lost in the snowstorm of output and obscure error messages. I'm just not sure at this point which mountain to move - put a gui with filtering on top of FlightCrew, or reimplement those "important" checks inside QC (and add a GUI to allow running multiple checks at once).
CSS validation I don't recall ever coming up on this thread (beyond the wish for detecting unused styles). I vaguely recall some discussion in the Sigil forum but I never paid any notice to it. It would have to be an offline option so if you have a particular candidate tool I'm happy to give it a quick whirl. Though of course it may suffer the same pitfalls as those other tools above and again be something that would be best implemented in Sigil...