Split indexing into a 2 step process so track numbers/types become known in advance
Big code cleanup and renaming
Originally committed to SVN as r2955.
Rename lots of things, THIS BREAKS THE AEGISUB BUILD because of changed exported type and function names.
Fixed an uninitialized memory bug that would make it crash on unindexed audio tracks in mastroska.
Made ffms.h C-friendlier.
Exports the start time of an audio track in the audio properties.
Less signedness and type conversion warnings.
Originally committed to SVN as r2940.
Fix an old bug that would drop the first or second video frame
Greatly improved audio decoding by adding a small internal cache
Originally committed to SVN as r2894.
