Remove potential infinate loop when caller hung up before audio completed. Thanks to telapi.com for the patch

FS-4781 --resolve I added some more code to the jb for seq but I still think the timestamp is more reliable than the seq since we can't know if the rtp is passing over a media proxy who may preserve the timestamps but still rewrite the seq, its more likely the ts shows the more accurate order of the rtp stream

FS-4779 I don't have this labbed up to try so can you try this patch and repost logs if it doesn't work

libopenh323-dev is in squeeze and sid, but it doesn't look like it's
going to make it into wheezy at this point.

H323Plus claims to be a drop-in replacement, but this should probably
be tested thoroughly.

Note that mod_h323 is still in the avoid_mods list pending testing to
verify that it builds cleanly on each target.

Thanks-to: Steven Ayre <steveayre@gmail.com>

FS-4567 --resolve

always update state to CS_HANGUP for now, even if the channel is gone to handle race where we miss this update

only put session pointers in when its done from a codec that comes from a session and set it to the right one

FS-3473: --resolve do not use asm optimizations on newer broken apple compiler until we can figure out the correct fix for this