Fix instance_key dtype check rejecting object vs str string ids

[LucaMarconato]

AI-generated—manually reviewed

Since 65dc73e the check in validate_table_in_spatialdata raised whenever the dtypes differed and either side was string-like. With pandas>=3 the index of shapes read from parquet is str, while anndata reads string columns as object, so valid datasets (e.g. the mouse_liver sandbox dataset) failed to load. Only raise for a string vs non-string mismatch.

Add a network regression test on the real cells dataset, annotating cell_boundaries via the cell_id column.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.42%. Comparing base (ea93a37) to head (8dcbc09).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1270      +/-   ##
==========================================
+ Coverage   91.40%   91.42%   +0.01%     
==========================================
  Files          53       53              
  Lines        8381     8381              
==========================================
+ Hits         7661     7662       +1     
+ Misses        720      719       -1     

Files with missing lines	Coverage Δ
src/spatialdata/_core/spatialdata.py	93.46% <100.00%> (+0.09%)	⬆️

🚀 New features to boost your workflow:

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

[LucaMarconato]

No changelog entry (I'll label as release-ignore): this fixes an unreleased regression. The stricter check came in with 65dc73e (#1197), which isn't in any release yet. The latest release, v0.8.0, still has the old dtype is str comparison, and that comparison never fires because a dtype object is never the builtin str. So released versions never raised this error.