Give samples with quick sampling->submission a relaxed upper clock filter limit, tighten upper limit otherwise

[corneliusroemer]

If the clock deviation is high, but the sampling was close to submission, it’s maybe not a bad sequence. Whereas high clock deviation and long time between sampling and submission is indicative of messup.

We could tighten our clock deviation that way, having an exception for rapidly submitted sequences. This way we wouldn’t throw away new interesting developments, like a sudden burst in mutations. That would show because it would be covered by the rapid submission exception.