Fixed: incorrect waiting for timeout in blocking-but-timeout mode send/receive functions

[ethouris]

This fixes the bug that has been introduced with timeout waiting in blocking mode with timeout setup for sending and receiving.

BEFORE: If the timeout was set up, then the CV waiting function was waiting for the time set up as timeout since the moment when the call is to be made - as it was in the while loop, it means that in case of spurious signals (or meaning something else), which caused the conditions to be re-checked and possibly waiting loop re-entered, the next waiting call started to wait again for the time of timeout since the moment the loop was reentered, not since the moment when the sending/receiving function was called.

EXAMPLE PROBLEM:

1. The function enters at 10:12:16.000500
2. Timeout is 2 seconds.
3. The data are not ready, the function enters the while loop and waiting on a CV up to timeout=2s
4. After 1.5 second the signal was issued on this CV.
5. The signal happens to be spurious - at least the "data ready or connection broken" condition is not satisfied. The loop is then continued.
6. By continuing the loop, the waiting on a CV is done, again, up to timeout=2s

Result: The function is waiting this time for 3.5s, while according to the configuration it should return with EAGAIN after 2s.

FIX: In point 6, waiting is done not for a relative time, but up to the given fixed point in time, which is calculated in the beginning as the function entrance time plus the configured timeout. So, in this case, as the fixed timeout timepoint is precalculated, in this case it would be distant by 0.5s from now, so it will exit after 0.5s, effectively 2s since the function entrance.

[maxsharabayko]

Yes, all good.
wait_until() returns false on timeout, and true if signaled or spurious.
On timeout can break the while loop due to the expiry time is reached.