Postgres: Handle negative replication delay

[Tenzer]

In case the master and slave's clocks are slightly out of sync, and there is virtually no replication delay, we may experience that a negative timedelta in the replication delay check. That is however not detected in the current code, as Python sets the days property to -1 and seconds to 86400, in order to not have a negative number in the seconds property.

This change checks if days is a negative number, and if it is, we always send 0 seconds as the replication delay.

There's an example of how it can look if it isn't fixed here:

[remh]

Thanks @Tenzer ! It looks good to me, i'll make a few tests and merge it really soon!

[remh]

Thanks!