Create script to periodically update mplayer to latest subversion, then re-test each test case uploaded. For each test case, three outcomes:
1 - test case repro-tested, still has bug
2 - test case not yet repro-tested
3 - test case repro-tested, bug disappeared
Here "bug" means comparing on bug type only, not on stack hash (since stack hash changes).
Fix Metafuzz index to report test case outcomes, most recent version of mplayer tried, with color codes. Green - repro'd, yellow - not yet tested, red - bug disappeared.