RTP: support payloads with larger packetization interval.

RFC 3551 section 4.2 said that a receiver should accept packets
representing between 0 and 200ms of audio data. Now we add the
ability to decode multiple frames in a payload as long as the
jitter buffer is not full. This change covers G711, GSM, and
GSM-EFR. AMR will be added later.

Bug: 3029736
Change-Id: Ifd194596766d14f02177925c58432cd620e44dd7