refactor(tests): testifylint don't use Equal for floating point tests

[Joibel]

Part of the refactor started by #13365 to enable us to use testifylint, shift to using InEpsilon or InDelta for floating point comparisons. Mostly use InEpsilon as it is more resilient, but using InDelta for testing against zero.

Reviewers: As with the others in this series, please feel free to apply patches as you see fit.

[agilgur5]

Code LGTM, but one question: Does testifylint require this? For general floating point arithmetic, this makes sense, but it seems like all of the cases here do not involve large or long floats and so don't have the same chance of being inaccurate (and the tests passed before too)

[agilgur5]

Mmm I'm going to suggest the same as I did in #13406 (comment) that auto-merge before approval is an anti-pattern. I didn't even notice it was enabled here

[Joibel]

Code LGTM, but one question: Does testifylint require this? For general floating point arithmetic, this makes sense, but it seems like all of the cases here do not involve large or long floats and so don't have the same chance of being inaccurate (and the tests passed before too)

Yes it does. Well, it is a default enabled linter which I agree with. It seems unlikely that rounding will affect the tests in question, but it felt better to leave the linter on.

Mmm I'm going to suggest the same as I did in #13406 (comment) that auto-merge before approval is an anti-pattern. I didn't even notice it was enabled here

Sorry, I missed that comment. I'll try to remember we're not using that here in future.

[Joibel]

Also worth noting that most of these are removed by #13265 where I've moved metrics which make most sense as integers to be integers.

[agilgur5]

Well, it is a default enabled linter which I agree with. It seems unlikely that rounding will affect the tests in question, but it felt better to leave the linter on.

Yea I prefer to stick with the defaults generally as well (for the same reasons that go fmt, prettier, and standard exist), just wanted to make sure.

Sorry, I missed that comment. I'll try to remember we're not using that here in future.

It's main use-case is to auto-merge once CI/tests are complete, so I think it only makes sense to use after a PR has been approved such that all changes (including to title and co-authors) are already in.