Add errors, to MagneticWell and MercierStability objectives if grid contains axis.

[dpanici]

Resolves #2097 by ensuring the situation (which is a bit nonsensical anyways) is not reachable by the user.

Enforces what we already had assumed which was that the user should never pass an on-axis grid point to the MagneticWell or MercierStability objectives (the latter bc it is undefined and should be NaN, the former because it is just zero on-axis anyways by definition of its limit, and also we get NaN in the reverse mode gradient of the on-axis point)

[github-actions]

Memory benchmark result

|               Test Name                |      %Δ      |    Master (MB)     |      PR (MB)       |    Δ (MB)    |    Time PR (s)     |  Time Master (s)   |
| -------------------------------------- | ------------ | ------------------ | ------------------ | ------------ | ------------------ | ------------------ |
  test_objective_jac_w7x                 |    6.15 %    |     3.824e+03      |     4.059e+03      |    235.11    |       39.56        |       36.64        |
  test_proximal_jac_w7x_with_eq_update   |    0.13 %    |     6.655e+03      |     6.664e+03      |     8.86     |       165.25       |       162.83       |
  test_proximal_freeb_jac                |   -0.22 %    |     1.323e+04      |     1.320e+04      |    -29.54    |       84.41        |       84.21        |
  test_proximal_freeb_jac_blocked        |   -0.72 %    |     7.559e+03      |     7.505e+03      |    -54.43    |       74.79        |       72.45        |
  test_proximal_freeb_jac_batched        |    0.62 %    |     7.454e+03      |     7.501e+03      |    46.36     |       73.57        |       72.56        |
  test_proximal_jac_ripple               |   -1.35 %    |     3.509e+03      |     3.462e+03      |    -47.24    |       68.07        |       64.45        |
  test_proximal_jac_ripple_bounce1d      |   -1.16 %    |     3.573e+03      |     3.531e+03      |    -41.40    |       76.42        |       76.51        |
  test_eq_solve                          |   -4.15 %    |     2.091e+03      |     2.004e+03      |    -86.77    |       94.73        |       93.91        |

For the memory plots, go to the summary of Memory Benchmarks workflow and download the artifact.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.52%. Comparing base (29c7877) to head (3d32db5).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2098      +/-   ##
==========================================
- Coverage   94.52%   94.52%   -0.01%     
==========================================
  Files         102      102              
  Lines       28785    28786       +1     
==========================================
  Hits        27210    27210              
- Misses       1575     1576       +1     

Files with missing lines	Coverage Δ
desc/compute/_stability.py	99.17% <100.00%> (ø)
desc/objectives/_stability.py	99.26% <100.00%> (+<0.01%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[ddudt]

I would prefer if we are able to correctly implement the axis limit, since it should be well defined and always 0.

[YigitElma]

• replace the division by 0 with safediv