Fix rotl90, rotr90 and rot180 for LocalOperator

[pbrehmer]

I stumbled upon this bug when I was trying to run SU for Hamiltonians with non-uniform physical spaces. It turns out that the rotation functions for LocalOperators would only work for $1\times1$ unit cells or uniform lattices - this should fix it. (I hope my cold-ridden brain actually did the right thing to fix it but I added some tests so it should be fine, hopefully.) I also renamed the functions for index rotations to remove the type piracy.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/algorithms/correlators.jl	100.00% <100.00%> (ø)
src/operators/localoperator.jl	88.70% <100.00%> (+1.09%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[lkdvos]

Looks good to me, left some minor comments.
If we are changing the name, I would also be happy to do something like: rotate_left_90, rotate_right_90, rotate_180, and use that everywhere, but obviously that's not a requirement.

[pbrehmer]

If we are changing the name, I would also be happy to do something like: rotate_left_90, rotate_right_90, rotate_180, and use that everywhere, but obviously that's not a requirement.

If we really want to change that everywhere, then I'd do that in a separate PR since we use rotl90 and rotr90 a lot throughout the code. In principle I would be open for that but I'm also fine with keeping a separate function for the CartesianIndex rotations. Reasons being:

• Using rotl90 and rotr90 for all other purposes makes a lot of sense matching the logic from Base and we want to avoid type piracy
• The CartesianIndex rotations require a different function signature where the unit cell size needs to be supplied so I feel that this is in spirit slightly different than the other rotl90 and rotr90 methods