Feature/3099 nan tolerance

[piotrb5e3]

Hi,

This is a proposed solution to #3099.

I unified the code of all tolerance matchers and improved handling around NaNs.

It has one issue: since the new code operates all on doubles, the old "Match close enough numbers" test fails, as the ranges of acceptable values are slightly different with double rounding. Not really sure what to do about it.

I'd also happily remove FloatToleranceMatcher, but I'm not sure if it's not part of the API.

[LeoColman]

Interesting solution by using Number!

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[sksamuel]

Just need to fix the test @piotrb5e3 and I think we're good!

[Kantis]

In my opinion we should revert the unification of ToleranceMatchers. It'll lead to some duplicate code, but it will let us be as correct as possible which I think is more important.

[sksamuel]

I agree

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[LeoColman]

@piotrb5e3 Could you give maintainers access to your branch so we can fix the git history?

[piotrb5e3]

@LeoColman Should I add specific people to the access list for this branch? Allow edits by maintainers is already enabled.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.