Fix for #1881, ProbabilitySamplingScheduler seems to be inverted

[DanBlackwell]

This should fix #1881. I have removed the check for 0.0 as I think that is a valid score potentially, if for some reason you really don't want to fuzz an input. Additionally I am not setting the parent ID in the on_replace function as it may be possible you replace a corpus entry with an input generated based on itself (though this may be somewhat weird, but maybe you're trying to minimise the length of all inputs as you go or something...)

[tokatoka]

yeah looks good. can you run cargo +nightly fmt and then we merge

[DanBlackwell]

Ok, have ran that and it reordered one import. Just waiting on CI now.

[DanBlackwell]

I've also added a debug_assert as I think that negative and infinite values would break the scheduler.

[tokatoka]

looks good i'll merge once it turns green