Mark node as not down when alive connection was acquired

[a-chernykh]

I've discovered this problem when running latest moped:master with thin server in threaded mode (--threaded). We have one Mongo server and when it goes down for a few secs, some of our app servers are unable to reconnect to it. We see lots of log output like this:

MOPED: Retrying connection attempt 1 more time(s), nodes is [], seeds are MOPED: Retrying connection attempt 1 more time(s), nodes is [], seeds are MOPED: Retrying connection attempt 1 more time(s), nodes is [], seeds are

[Nothus]

:+1:

[a-chernykh]

@arthurnn @durran could you review this please?