Merge pull request #4698 from RosettaCommons/vmullig/tweak_simple_cycpep_predict_sci_test_threshold
Tweak thresholds on simple_cycpep_predict scientific test a bit.
The recent weak to the Dunbrack libraries seems to have made simple_cycpep_predict a little bit slower (less than 10%). As such, the simple_cycpep_predict scientific test frequently fails just because it isn't sampling as much as expected. This PR relaxes the sampling threshold just a bit, since I think we'll just have to accept the slight slowdown.