Excised fluxes considered harmful

[bprather]

I've now gotten three separate reports from users with boundaries/excise_polar_fluxes=true that their simulations have proven unstable. I'm not certain right now that the option is at fault, but I do strongly suspect it and we're working to test directly.

I suspect the issue may be fundamental to the algorithm (recall we are using a sort of "H-AMR lite," not a published scheme). I have not reproduced the issue, but my suspicion is that in directly connecting cell centers, the scheme is somehow superliminal with respect to the usual timestep calculation (even accounting for half-sized zones).

Since it may not be fixed soon, and may be invasive to the timestep calculation when it is fixed, we disable it by default with a cautionary note here, and we can drop the fixed or improved scheme into dev later.

Mea culpa, enabling the shiny untested toys as defaults again. When will I learn?