fix: don't mark missing multiscales metadata as processed

[wietzesuijker]

Problem

add_multiscales_metadata_to_parent() can legitimately return None when inputs are insufficient to generate multiscales metadata, e.g.

• fewer than 2 resolutions are available
• CRS is missing
• bounds are missing

Previously, this “skipped” outcome was still recorded in processed_groups["/measurements/reflectance"], making runs look complete even though no multiscales metadata was written.

Fix

• Type add_multiscales_metadata_to_parent() as xr.DataTree | None.
• If it returns None, log an info message and do not record it in processed_groups.
• Preserve parent-group attrs during consolidation by avoiding recreation of groups that already exist on disk (prevents overwriting attrs like multiscales / zarr_conventions).
• Make multiscales metadata writing more robust when .rio.* metadata is unavailable (fixture-like inputs):
  • prefer rioxarray for CRS/bounds when available
  • fall back to provided crs
  • fall back to estimating bounds from 1D x/y coords when .rio.bounds() is unavailable

Notes

• Multiscales metadata is written as parent-group attributes (no dataset write).
• processed_groups intentionally tracks leaf dataset groups written (used later for consolidation), so it should not be updated for the attrs-only step.

Testing

• uv run pytest -q tests/test_s2_multiscale.py tests/test_s2_multiscale_geo_metadata.py
• uv run pytest -q tests/test_cli_e2e.py::test_convert_s2_optimized

[emmanuelmathot]

Here processed_groups is never updated...

[wietzesuijker]

Yes, that's intentional, @emmanuelmathot. I added 2b43aa0 to clarify why:

multiscales metadata is written as parent-group attrs (no dataset write). processed_groups is used to track dataset groups written (and is later used for consolidation), so it is intentionally not updated for the attrs-only step.

The naming is perhaps not the easiest here. I added some suggestions to the PR description above

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 61.81818% with 21 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/eopf_geozarr/s2_optimization/s2_multiscale.py	39.39%	20 Missing ⚠️
...pf_geozarr/data_api/geozarr/multiscales/geozarr.py	92.85%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!