Polar bug redux

[bprather]

I've now gotten three separate reports from users that various simulations have proven unstable, usually originating at the pole. Based on my reading of the reports there seems to be a common theme of using excise_polar_flux, and the affected versions seem to be 2024.9 and up. Please feel free to post plots or diagnostics in this issue, because I am having a very hard time reproducing the issue on any of my builds or hardware.

Given the common theme, I'd first advise disabling excise_polar_flux -- and if that doesn't work, revert to 2024.5.1 unless you need the new features in 2024.9 (which included very little for ideal GRMHD except excise_polar_flux). If simulations fail with 2024.5.1 please let me know, since I'm working under the assumption this is a 2024.9 thing.
Two other mitigations to try, if you need excise_polar_flux to stay on or continue to experience the issue in 2024.5.1:

1. Lowering the CFL condition, in case the issue is timestep related
2. Using the imex driver, in case the issue is related to boundary conditions (i.e., WIP: Stop running unnecessary UtoP at boundaries #128)

If either of these helps I'd be very interested to know.

Currently the PR #126 just disables excise_polar_flux by default with warnings, but any real work toward fixing this thing should end up there, i.e. in the branch fix/excise-flux-bad

[bprather]

Oops, one fix does not mean the fix. Leaving this open until we can either confirm the race condition fix eliminates issues, or find the rest of the gremlins responsible.