`TestMarginalVsMarginalApprox` is now flaky

[ricardoV94]

The test has failed in both #6062 and #6067

I never saw it fail before, so I presume it's due to changes introduced in #6055. It should be enough to set the jitter manually to 0 as it was before. Is this reasonable?

CC @danhphan @bwengals

[danhphan]

Yes, thanks @ricardoV94, let me check it in detail.

[bwengals]

Ah thanks guys. It's reasonable in the short term. Long term these tests need to be updated alongside a fix for #5922. It might be best to just mark these as xfail in the meantime.

[ricardoV94]

It might be best to just mark these as xfail in the meantime.

We have a strict xpass, so our tests would fail whenever something marked as xfail actually passes.

[danhphan]

Thanks @bwengals for the fix :)

[bwengals]

Sure @danhphan, but please take a look. I think the issue was just the sensitivity, but I know you've been dealing with jitter issues lately so would still appreciate if you checked #6076 over