The logic in cc_post_recovery() actually has two parts: pre_post_recovery logic and post_recovery logic. So introduce this split.
Improves each individual cc's responsibility for congestion control. As each
registered .cong_signal() enters RECOVERY, so each registered
.post_recovery() exits RECOVERY from each cc module in an organized place.
This improves code reading.
Currently, each cc module shares the same post_recovery() function, but that
is not necessary a MUST. This allows later individual cc improvement.
No impact to rack and bbr.