Vega Strike Code

Bugfix: playtime query on streaming sources was broken:
OpenAL only reports the current buffer's play time, so in order to get accurate time stamps,
we must add the current buffer's start time to the reported time.

Now the tests report 27ms maximum drift: ain't that great?

move ffmpeg module specific advance flags back to ffmpeg module, GTK2 still needs this done too

Merging fix from safemode branch -r12759, forgot to cache a few variables

cmake wasn't caching a few variables that we needed to

merging to safemode branch -r12757, fixes to cmake FFMPEG detection and general ffmpeg related cmake cleanups

cmake cleanups (mostly ffmpeg related)

Merging -r12749 of FindFFMPEG.cmake from audio branch to safemode branch, fixes detection for distro installed headers and libs

Merging trunk -r12745

Ticket #37: Integrate SUCSS part 1, initialize with borrowed AL context

Lots of fixes and some new tests:
- Streaming fixes to account for broken OpenAL implementations that don't support setting AL_SEC_OFFSET
- Streaming tests now measure time drift. I'm measuring some weird jitter, though... that might be a problem with movie playback
- Complex scene test now uses a quicker tick, to test the scene manager's efficiency with many sources - which seems good for now :-D
- New test for streaming the same track twice - concurrently. Will be interesting to know if Windows can handle that (windows has some weird file sharing rules - ie: opening a file twice).
o While doing that, I noticed we didn't have a generic API for playing a stream from a position other than the beginning. Fixed that.
- Warning fixes (initialization order)