Possibility of infinite loop in jitter buffer (thanks Michael Broughton for the report).

[nanang]

The full report can be found ​in this thread.

The problem seems to be caused by modification in ticket #744.

The corresponding ticket for 1.0.x branch is ticket #763

[nanang]

In r2578:

• Updated loop condition in put_frame() to avoid possibility of infinite loop.
• Added JB capabilities to handle sequence restart & jump.
• Updated jitter calculation, e.g: reset max_hist_level after updating prefetch, avoid updating prefetch when burst level is exceeding max_burst.
• Updated shrinking method to be less agressive (only shrink JB when JB size is twice larger than burst level).
• Updated the way JB switching status from 'initializing' to 'processing' by waiting for some OP switch cycles.
• Few simplifications in framelist process, e.g: replacing fields 'empty' & 'tail' with 'size'.
• Minor updates: comments, shortened framelist field names, added some JB states for reporting/monitoring purpose.

[nanang]

In r2579, updated JB test:

• Added support for more scenarios, e.g: frame loss, late, sequence restart, etc.
• Added automatic test result checks based on specified condition.
• Updated test data.

[nanang]

In r2583:

• Fixed jbuf in handling far sequence jump (the distance is over jbuf capacity) while jbuf is empty.

[nanang]

In r2585:

• Updated delay calculation method & added statistic report in jbuf_destroy().

[nanang]

In r2587:

• Replaced stricmp() with pj_ansi_stricmp() in jbuf_test.c for cross platform compatibility.

[nanang]

In r2662:

• Cleaned up jbuf.c of unused function jb_framelist_origin().

[nanang]

In r2672:

• Fixed jitter buffer in handling DTX with relative short period of 'keep alive', e.g: Speex DTX seems to periodically send a keep alive frame every 20 silence frames.