[mansurjisan]

Description

Fixes on top of #74 (feature/pacific) so the Pacific ops features do not change the systems that already run (STOFS-3D-ATL, SECOFS, SECOFS-WW3, STOFS-3D-AK), while keeping the Pacific settings. For merging into feature/pacific; please use "Create a merge commit" (not squash), so the nos-workflow #238 submodule pointer to this commit stays valid.

What #74 changed for existing systems, and the fix:

• ATL ADT datum offset cancelled (+0.45 m on elev2D versus ops). The -0.45 m was moved to an in-memory copy of ssh, where it cancels in ssh(t) - ssh(0). Restored per day in _regrid_esmf / _read_adt (float32 before the mean, as main), and _apply_adt_blend writes the real ssh variable again. The Pacific uses its own -0.4715517 through the same path. from_yaml now defaults adt.datum_offset to -0.45 like the dataclass.
• ATL TSUV_1.nc lost depth. In _stofs_prepare_tsuv a one-file group left ds unassigned; the exception was swallowed, so depth was never written and the datasets were never closed. The Fortran OBC and nudging tools read level depths from it. Fixed with a finally close; TSUV_1.nc matches main bit for bit again.
• ATL RTOFS slots 23 -> 22. New obc_min_slots (default 23; yaml obc.min_slots); obc_min_timesteps is the QC threshold again. The Pacific sets 4.
• OPS_OBC_DT NameError on the opt-in Fortran path (rtofs.py): uses config.obc_dt.
• ADT region: the fixed ops ROI again for -180/180 systems; the SSH_1 extent for the 0-360 Pacific.
• ATL rerun archive: back to main's four entries; temnu/salnu/surface restore/elev2dth only with the new obc_archive_extended (yaml obc.archive_extended), set by the Pacific presets.
• SECOFS, SECOFS-WW3, STOFS-3D-AK nudging 3 h -> 6 h. nudging_dt defaults to 10800 again (their param.nml keep step_nu_tr = 10800); the ATL ops timeline stays 21600; the Pacific sets 86400 explicitly.
• Surface restore: off in the presets and from_yaml; process() refuses to run unless NOS_SURFACE_RESTORE_EXPERIMENTAL=1 (it reads T and S from one file though the daily files split them, uses the boundary nodes, and has a placeholder time axis). The code is kept; it needed a find_input_files to be instantiable.
• Pacific values: forecast_hours 48 (ops rnday 3.0 = 24 h nowcast + 48 h forecast); elevation segments [0, 1, 9] unchanged.
• from_yaml tolerates null forcing.ocean, nudging and surface_restore blocks again.

Guard tests (tests/test_stofs_config.py): ATL offset -0.45 with and without the yaml key, 23 slots, QC 21, obc_dt 21600, archive off, surface restore off; SECOFS/SECOFS-UFS/unspecified yaml nudging 10800; the ATL archive map; from_yaml parsing of the new keys; _stofs_prepare_tsuv run unmocked keeps depth and logs no failure; the Pacific values.

Still for the Pacific (not changed here): obc_dt, obc_min_slots and the daily_3z discovery only act under the ops timeline, which the Pacific yaml leaves off; the TSUV_1.nc time axis is fixed at 21600 s; the ADT method and the smoothing time axis need checking against ops; surface restore needs a rewrite before it is enabled. The branch is behind main (#75-#79) and should take main before #74 merges.

Type of change

• Bug fix (non-breaking change that fixes an issue)

Change characteristics

• Changes outputs (values, COMOUT files or names, archives)? NO relative to main: ATL, SECOFS and AK outputs return to main's. YES relative to the current Enable STOFS Pacific System #74 head, which this corrects.
• Breaking change to existing functionality? NO
• Requires a nos-utils submodule bump (nos-workflow) or a paired nos-workflow change (nos-utils)? YES: nos-workflow #238 bumps its pointer to this commit (paired PR into feature/pacific_mvp).
• Requires new or regenerated fix files on the machine? NO
• Requires a re-prep or a new hotstart before the next cycle? NO
• Machines affected: both

How has this been tested?

• Full nos-utils suite: 1289 passed, 9 skipped (the Enable STOFS Pacific System #74 head had 31 failures; 1309 on main, the difference is two test files main has and this branch does not yet).
• Every preset and every nos-workflow yaml resolves to the same configuration as on main outside the Pacific (only new fields, at defaults that keep the old behaviour).
• TSUV_1.nc for a synthetic ATL case matches main bit for bit.
• Not yet run on a machine.

Checklist

• Dependent PRs merged and published (nos-utils before any submodule bump)
• pytest suite passes
• Existing operational behavior preserved, or the change is called out above
• No edits to a shared FIXofs while it is cycling
• Self-review done; yaml comments and docstrings updated where behavior changed
• No markdown files added to the repo (this template is the only exception)