Rescale bubbly volume fractions per cell in pre_process #2074

With mpp_lim and bubbles_euler, s_assign_patch_species_primitive_variables
rescaled the liquid volume fractions with whole-array operations on
alf_sum%sf (0:m,0:n,0:p) and the ghost-padded q_prim_vf fields from inside
a per-cell routine. The shapes differ, so this read and wrote out of
bounds (gfortran: "Array bound mismatch ... (308/300)"), and it rescaled
the entire domain once per cell.

Replace the five copies with s_scale_volume_fractions_to_void, which
rescales only cell (j,k,l) with a scalar sum, and drop the module-level
alf_sum array (which was never deallocated).

Add test E5ACA113 (1D -> Bubbles -> 2 Fluid(s) -> mpp_lim), which aborted
in pre_process before this change. The 64 existing Bubbles tests pass
unchanged.

A fix for this is already written. The pull request was closed only to reduce the number of simultaneously open PRs from this author, not because the work was rejected or found incorrect.

• Closed PR: Rescale bubbly volume fractions per cell in pre_process #2002
• Branch: fix/bubbles-mpp-lim-assign on the sbryngelson fork, at 6dacec272d
• Reopening needs no force-push and no rebase: gh pr reopen 2002 -R MFlowCode/MFC

Tracked with the pr-to-do label so the whole set is discoverable together.

Filed with Claude Code.