libsrtp: wraparound with loss decryption failure

[ibc]

Rationale in the ticket: https://webrtc-review.googlesource.com/c/src/+/358360

Theoretical solution is that the first output seq number of each mediasoup XxxxConsumer is not 0 or 1 (to avoid that a previous packet arriving later triggers the bug in libsrtp) and it's smaller than 2^15.

[fippo]

does your sequence number mapper handle this case "correctly":

• map(5) -> 5
• map(4) -> 4 I always assumed one would not have to bother with 4 in this case since it is too old but...

[ibc]

does your sequence number mapper handle this case "correctly":

• map(5) -> 5

• map(4) -> 4

I always assumed one would not have to bother with 4 in this case since it is too old but...

There are like 10000 tests covering that case and more complex ones in TestSeqManager.cpp.