Fix nextLevel quality switch

[mikrohard]

Description of the Changes

nextLevel quality switch is broken since 4a68b55
maxStartPTS is never calculated and thus the buffer is not flushed which makes nextLevel switch essentially the same as loadLevel

CheckLists

• changes have been done against master branch, and PR does not conflict
• no commits have been done in dist folder (we will take care of updating it)
• new unit / functional tests have been added (whenever applicable)
• Travis tests are passing (or test results are not worse than on master branch :))
• API or design changes are documented in API.md

[mikrohard]

I have also an additional question regarding this. Once I managed to fix the nextLevel quality switch the streaming occasionally stopped during the quality switch. What happened was that two of the three segments (LIVE HLS playlist) were looping for a few seconds until the playback stopped. After some debugging I found out that the _findFragment was returning the last two fragments to load (e.g. 2,3,2,3,2,3, ...) because it wasn't able to find the correct PTS to continue. I noticed that in all of those cases the deltaPTS value was huge (15-20 seconds) while it is normally around 2 seconds. If I add the fallback to check the huge deltaPTS and in this case load the previous fragment instead of the next one it solved all the problems. But it feels like a workaround to me. Any thoughts?

[mangui]

well spotted thanks
regarding your last comment this deserves a separate ticket with log