Fix floating point equality comparison in layouts

[penelopeysm]

Description

Fixes #4971 by using an approximate equality check.

Attribution

• I am listed in .zenodo.json (see How to get acknowledged #3503)

Things to consider

• Does it work on log scales?
• Does it work in layouts?
• Does it work in recipes?
• Does it work with multiple series in one call?
• PR includes or updates tests?
• PR includes or updates documentation?

[penelopeysm]

Couple of questions for anyone reviewing:

1. Although I haven't managed to find a series of floats for which the sum is more than eps() away from 1, I'm still worried it's a bit too conservative. Would something like eps() * length(heights), or eps() * sqrt(length(heights)) be a more sensible tolerance? I don't have a theoretical background in numerical computing :)
2. Do you want tests for these lines of code? Happy to add them if you do.
3. If the PR is accepted, do you want me to port this to the v2 branch as well?

[BeastyBlacksmith]

Couple of questions for anyone reviewing:

1. Although I haven't managed to find a series of floats for which the sum is more than eps() away from 1, I'm still worried it's a bit too conservative. Would something like eps() * length(heights), or eps() * sqrt(length(heights)) be a more sensible tolerance? I don't have a theoretical background in numerical computing :)

IIRC isapprox compares the first half of the bits to be equal, that should be sufficient in this case, I think.

2. Do you want tests for these lines of code? Happy to add them if you do.

Yes, please

3. If the PR is accepted, do you want me to port this to the v2 branch as well?

Yes

Thanks for working on this ❤️

[penelopeysm]

@BeastyBlacksmith Okay, I think this should be ready!

CI failures seem to be unrelated, they're occurring due to a bump in Latexify from 0.16.4 to 0.16.5, specifically this PR korsbo/Latexify.jl#298.

I don't mind fixing that in another PR :)