Fix ContainsGroupError when writing a DataTree with region/append_dim

[aaronspring]

Problem

Writing a virtual DataTree with region or append_dim (forwarded to each node) always failed with zarr.errors.ContainsGroupError. virtual_datatree_to_icechunk created every node's group unconditionally with Group.from_store, but region/append_dim require the group and its arrays to already exist — so the write blew up before it ever got to writing the region/append.

This is the same class of failure as #1001, but for the DataTree path.

Fix

Mirror the branch that virtual_dataset_to_icechunk already had on main: open the existing group for region/append_dim, create it otherwise.

for store_path, vds in paths_and_virtual_datasets:
    if kwargs.get("append_dim") or kwargs.get("region"):
        group = Group.open(store=store_path, zarr_format=3)   # must already exist
    else:
        group = Group.from_store(store=store_path, zarr_format=3)  # create

Test

test_write_datatree_region was marked xfail with the reason "This doesn't work in xarray either, maybe not even intended?" — but that was a misdiagnosis. On main, running it with --runxfail fails with ContainsGroupError at the group-creation step, not in any xarray region logic. With this fix the test runs to completion and its assertions pass, so the xfail marker is removed.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.33%. Comparing base (be604ed) to head (148ea28).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1036   +/-   ##
=======================================
  Coverage   90.32%   90.33%           
=======================================
  Files          37       37           
  Lines        2233     2235    +2     
=======================================
+ Hits         2017     2019    +2     
  Misses        216      216           

Files with missing lines	Coverage Δ
virtualizarr/writers/icechunk.py	91.66% <100.00%> (+0.10%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[aaronspring]

CI note: the 3 red build jobs (test-py312, test-py313, minimum-versions) fail only on test_parsers/test_grib.py (test_grib_virtual_dataset, test_grib_only_variables, test_grib_chunk_decodes_via_codec) — the GRIB/gribberish parser returning empty data_vars / a NodeTypeValidationError. These are pre-existing and unrelated to this PR, which touches only the icechunk DataTree writer; all 550 other tests pass. The same 3 failures appear on #1034 (a pre-commit.ci autoupdate that changes no source), #1032, #1033, and on scheduled main runs.

[TomNicholas]

Thanks @aaronspring !