Skip wasted cs-level table copy in render_shapes/render_labels

[timtreis]

Summary

Closes #415. _render_shapes and _render_labels were calling filter_by_coordinate_system(filter_tables=True), which copies the table's sparse .X via _filter_table_by_element_names. The filtered table was then immediately discarded — overwritten by join_spatialelement_table (shapes) or by per-element subset + match_table_to_element (labels via _set_color_source_vec).

For the multi-sample Visium use case in #415 (49 elements sharing one ~200k-row sparse table) this copy dominated the render call. The micro-benchmark in the issue investigation shows ~5.2× speedup on a synthetic 98k×3k sparse table; the gap scales with table size.

Switching both call sites to filter_tables=False is safe because:

• sdata_filt[table_name] is never read between the filter call and the point where it is overwritten (verified by grep + walk of render.py).
• sdata_filt.tables is never iterated by any code on the affected paths.
• Auto-resolution helpers in utils.py that iterate sdata.tables are only triggered when table_name is None, which already used filter_tables=False.

Unintended side effect, caught & fixed

_filter_table_by_element_names mutates sdata's table obs in place via table.obs = pd.DataFrame(table.obs), which triggers AnnData's implicit int64→str index coercion. Removing the cs filter exposes AnnData's strict _normalize_index rejection of non-RangeIndex int64 obs indices in the EntityID-collision workaround path (test_render_shapes_color_with_conflicting_index_name).

The fix extends the existing workaround in _render_shapes (originally added for scverse/spatialdata#1099): when a name collision is detected, also swap the obs index to a fresh RangeIndex, saved and restored under try/finally.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 87.50000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 77.81%. Comparing base (8a6a33f) to head (bb4c2e2).

Files with missing lines	Patch %	Lines
src/spatialdata_plot/pl/render.py	87.50%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #680      +/-   ##
==========================================
+ Coverage   77.79%   77.81%   +0.01%     
==========================================
  Files          11       11              
  Lines        3693     3700       +7     
  Branches      877      879       +2     
==========================================
+ Hits         2873     2879       +6     
- Misses        490      491       +1     
  Partials      330      330              

Files with missing lines	Coverage Δ
src/spatialdata_plot/pl/render.py	86.98% <87.50%> (-0.13%)	⬇️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

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