A problem with ASYNC drain

[mmacy]

Summary

If a thread is running async-drain (like TCP does) its possible some other thread may attempt to restart
the timer. If that occurs the thread should be given a error return and the timer not started (just like
the synchronous drain).

Note that Gleb's previous patch to the callout system fixes 1/2 of the problem this rarer case is still possible and
needs to be addressed :-)

Test Plan

Run it in a netflix work load as well as create a test module to create the exact circumstance.

[rrs]

missed an unlock

[rrs]

Originally I had started working on this before Gleb decided to take over and clean it up. I
had fixed the three issues I saw:

1. Principle - the start of a new timer was not rejected (like it is for sync drain) for async-drain.
2. A three way leak in the callout start
3. An incorrect return on some cases with async_drain and tcp.

I think these were the real issues that Gleb went in to fix when he re-wrote a block of
the code and started breaking thing due to the change in return codes and others.

I decided to go back to this version, which I know works and NF is testing (and will soon release
into production).

If Gleb wants he can hack on his re-write and figure out how to make it work.. or he can use this
here and it will work.. I have no preference either way.. I just know that this code does work :-)

[mmacy]

Can I close this?

[jtl]

Yes, see rS302894, rS302997, rS302998, rS303037, rS303096, and rS303098.

Short version: @glebius eventually fixed it.