Updated version of "Move parse_tracers file from physics to host"

[grantfirl]

Original description/work from @dustinswales:

This PR moves the GFDL_parse_tracers file from the physics to the host (where it belongs)
Also, is a small bugfix in a plotting script.

@grantfirl update:

• remove parse_tracers module entirely since we can use a local version of get_tracer_index() already in SCM code
• move get_tracer_index() function from scm_type_defs to GFS_typedefs to avoid circular dependency

[grantfirl]

@mkavulich @scrasmussen @hertneky @dustinswales This introduces another "delta" between the fv3atm and SCM version of GFS_typedefs. The fv3atm version (PR to come) will use get_tracer_index() from FMS (which is not used in SCM) like the rest of fv3atm instead of the previously included version from GFDL_parse_tracers in ccpp-physics.

[grantfirl]

Replaces #520

[scrasmussen]

Looked over changes, get_tracer_index is an equivalent and slightly cleaner version to the one removed from the ccpp-physics repo. Looks good and compiles!

[grantfirl]

The same RTs from #566 are expected to change results here since new baselines haven't been uploaded to mohawk yet.

[grantfirl]

New baseline artifacts from https://github.com/NCAR/ccpp-scm/actions/runs/13975876703 need to be uploaded when @scrasmussen @hertneky @mkavulich gets a chance.