Disable checkIniTemp/checkIniSalt for repeat-year restart ICs

The model produces theta=0 and salt=0 at 63 wet cells on the northern
OBC boundary (j=423, mid-depth). These are legitimate model state, not
uninitialized data, so the INI_THETA check must be disabled for
pickup-derived initial conditions.

Also remove the NaN/Inf fill from pickup_to_init — the big-endian fix
eliminated the overflow, so the fill was no longer needed.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: fluidnumericsJoe

simulations/glorysv12-curvilinear/input/data

@@ -73,6 +73,6 @@
  hydrogSaltFile='S.init.bin',
  uVelInitFile='U.init.bin',
  vVelInitFile='V.init.bin',
-# checkIniTemp=.FALSE.
-# checkIniSalt=.FALSE.
+ checkIniTemp=.FALSE.,
+ checkIniSalt=.FALSE.,
  &end

spectre_utils/pickup_to_init.py

@@ -123,9 +123,7 @@ def pickup_to_init(pickup_prefix: str, output_dir: str, nx: int, ny: int, nr: in
             init_path = out / init_name
             if init_path.is_symlink() or init_path.exists():
                 init_path.unlink()
-            data32 = data.astype(">f4")
-            data32[~np.isfinite(data32)] = 0.0
-            data32.tofile(init_path)
+            data.astype(">f4").tofile(init_path)
             size_mb = init_path.stat().st_size / 1e6
             print(f"  Wrote {init_path} ({size_mb:.1f} MB)")