Add Benchmarks regarding Coil Optimization

[YigitElma]

• Adds benchmarks for compute and jacobian of QuadraticFlux with a mixed coilset that is complex enough.
  • I tried to hit NFP, sym, and different coilset for loops of compute_magnetic_field
  • These tests will start running properly once the PR is merged, because those actions run under pull_request_target, that means they always use master code
• Fixes bug in plot_coils and CoilSet.from_symmetry that ignored the given check_intersection
• Adds check_intersection to coil set initializers
• Sets the default of check_intersection to False in many places. Only exception is FourierCurrentPotentialField.to_CoilSet as it seems a bit more important.
• Passes built LinearConstraintProjection object in perturb benchmarks to make it faster. We internally use it in this way anyway. Also, we benchmark LinearConstraintProjection.build() separately. Also takes it out of setup to prevent triggering build multiple times.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.30%. Comparing base (f0b58b8) to head (ce1adf7).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2266      +/-   ##
==========================================
- Coverage   94.31%   94.30%   -0.01%     
==========================================
  Files         101      101              
  Lines       28932    28934       +2     
==========================================
- Hits        27288    27287       -1     
- Misses       1644     1647       +3     

Files with missing lines	Coverage Δ
desc/coils.py	97.46% <100.00%> (-0.51%)	⬇️
desc/plotting.py	95.41% <100.00%> (ø)

... and 3 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[YigitElma]

Default to no check for intersections, change the docstring for new default, add to the changelog.

[YigitElma]

Sorry for the last minute changes, I realized that memory benchmark script had a bug in the import and the benchmarks in general were taking too long on CI (the jacobian one was 4 seconds on my laptop CPU but 40 seconds on CI). I fixed the bug and made the tests smaller.