fix: extend patch validation loops to num_patches_max

[SVS87]

Description

The alteration rights, smoothing, and primitive variable validation loops were bounded by num_patches, making the inactive-patch else branches unreachable. Changed all three loop bounds to num_patches_max to match the geometry loop, so inactive patch slots are validated across all four passes.

Closes #1482

Type of change (delete unused ones)

• Bug fix

Testing

Ran ./mfc.sh test -j 4 locally on Ubuntu/WSL with GNU compiler and MPI. 578 tests passed, 0 failed. The fix is output-neutral for valid cases — the newly-reached inactive-patch checks only abort on misconfigured inputs that no passing test exercises.

AI code reviews

Reviews are not retriggered automatically. To request a review, comment on the PR:

• @claude full review — Claude full review (also triggers on PR open/reopen/ready)
• Or add label claude-full-review — Claude full review via label

[sbryngelson]

Traced this and it looks correct and complete. With the bound change, all four validation passes — geometry (already at num_patches_max), alteration rights, smoothing, and flow variables — now iterate over inactive slots too, and each routes them to its inactive/unsupported branch: s_check_inactive_patch_alteration_rights (L102), s_check_unsupported_patch_smoothing via the else at L113-114, and s_check_inactive_patch_primitive_variables (L123). patch_icpp is sized num_patches_max, so the widened loops are in-bounds, and for valid cases the inactive slots hold defaults so the new checks are output-neutral (consistent with the 578-test pass). LGTM.