[BUG] Main/backup: The activation period ends earlier than expected

[mbakholdina]

Describe the bug

During testing the RTT estimate convergence for Connection Bonding (main/backup algorithm), I've noticed that the activation period on the main link ends earlier than expected. It happens at the beginning of a transmission, the activation period ends almost right after 1 RTT is passed since the start of a transmission. As a result, the value of link stability timeout changes to dynamic that depends on the incorrect RTT estimate (actually the initial value of 100 ms). At this time, we haven't yet received the first ACK with the first RTT estimate.

Here the first ACK comes

There is also one problem with RTTVar estimate on the sender side, see ticket #1769.

To Reproduce

See issue #1755 for the reproduction steps.

Failed test case:

Test Case 3. Main link shutdown for 40 seconds (higher than SRTO_PEERIDLETIMEOUT = 5 s)

Scenario	Main Link (enp7s0)	Backup Link (enp8s0)	Bitrate, Mbps	SRT Options
-	RTT1 = 1000 ms (1 s)	RTT2 = 1000 ms (1 s)	1	latency=4000

Statistics and screenshots:
dlst_scen4_rtt1s_test3 #1768.zip