fix(templates): Use instance-suffixed MASKTABLE for AUTO_MASKTABLE runs

[johnpaulalex]

Changes default MASKTABLE for auto_masktable runs to be instance-suffixed, resolving file collisions for multi-instance runs (Issue #269).

Main changes:

• Adds MASKTABLE runtime parameter definition into param_templates/MOM_input.yaml, superseding the old fortran defaults. The non-AUTO_MASKTABLE default is effectively unchanged (still a single file shared across all instances), while the AUTO_MASKTABLE case is now defaulted to have an instance suffix, which fixes the file writing collisions.
• Adds unit tests in tests/check_buildnml.py verifying MASKTABLE expansion across single and multi-instance modes.
• Adds two multi-instance SMS_C2 system tests (with AUTO_MASKTABLE on and off) to cime_config/testdefs/testlist_mom.xml. Note the non-auto-masktable version won't find the static masktable file so it runs in unmasked mode.

This PR relies on NOAA-GFDL/MOM6#1181 to actually use the MASKTABLE parameter in the AUTO_MASKTABLE case.

Note the PRs can be submitted in either order:

• If this PR is submitted first, it will set the MASKTABLE parameter for AUTO_MASKTABLE runs, but MOM6 won't read it yet.
• If the MOM6 PR is submitted first, it will listen for the MASKTABLE parameter for standalone AUTO_MASKTABLE runs, but CESM won't set it yet.

Resolves #269 (once CESM points to a version of MOM6 with its PR live as well)

Testing:

• tests/check_default_params.py: PASS (yaml translates to json correctly)
• tests/check_buildnml.py: PASS
• aux_mom.derecho (System tests): both new C2 tests passed, see /glade/derecho/scratch/jpalex/cs.status.issue269_v4. (I also ran the rest of aux_mom, which ran fine but didn't do any baseline diffing - please lmk if there's an incantation to do that)

Google Antigravity was used to write the code and tests, followed by human-guided verification.