IssueTracker

MediaExtractor#readSampleData() reads 3 bytes shorter than I expect when it reads a text/3gpp-tt track.
https://stackoverflow.com/questions/58049727/how-to-retrieve-duration-of-3gpp-tt-timed-text-subtitle-or-closed-caption-via

Currently the result buffer contains:
- the first two bytes: the length of the rest (which is useless because readSampleData() returns bytes it has read)
- the rest: the timed text data

I want it to read three more bytes before them, which you see as SDUR in the RFC https://tools.ietf.org/html/rfc4396#section-4.7

Or I want MediaExtractor to provide another method MediaExtractor#getSampleDuration which returns a long microseconds.