fix(datashader): render_shapes in a non-global coordinate system (#782)

[timtreis]

Closes #782.

Problem

render_shapes(..., method="datashader") raises KeyError: 'global' when the shapes element has no "global" coordinate system (e.g. Visium HD data from spatialdata-io, which places elements in a custom CS). Datashader is used explicitly or automatically for >10k shapes, so this breaks the technology_visium_hd example notebooks.

Root cause

The datashader branch of _render_shapes wraps the already-transformed geometry with ShapesModel.parse(...) and then requests the canvas extent in a hardcoded "global" CS. Whether the GeoDataFrame constructor propagates the source element's .attrs is pandas/geopandas-version dependent:

• When attrs propagate (reporter's fresh stack), the wrapped element carries only the real CS (my_sample), so the "global" lookup in _element_extent_fast raises KeyError: 'global'.
• When attrs are dropped (older pandas), parse defaults to a global identity and it silently works.

Fix

In the datashader branch (src/spatialdata_plot/pl/render.py):

• Drop any carried-over transform attr before parsing (otherwise ShapesModel.parse with transformations= would raise ValueError: Transformations are both specified for the element and also passed as an argument).
• Tag the wrapped element with transformations={coordinate_system: Identity()} — the geometry is already in the target CS.
• Compute the extent in coordinate_system instead of the hardcoded "global".

This mirrors the already-correct points datashader branch. The working (global) case is byte-identical — same coordinates, same extent — so no visual baselines change.

Notes

• Verified the repro (blobs moved to a custom CS, rendered with method="datashader") and confirmed the fix holds in both attrs regimes (propagated / dropped).
• No test added per maintainer preference; the behavior cannot go red→green in the current local pandas (attrs dropped by the constructor), and test_plot_* are CI-only.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 80.10%. Comparing base (9f19b7d) to head (057bd5c).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #783   +/-   ##
=======================================
  Coverage   80.09%   80.10%           
=======================================
  Files          18       18           
  Lines        4863     4865    +2     
  Branches     1080     1080           
=======================================
+ Hits         3895     3897    +2     
  Misses        603      603           
  Partials      365      365           

Files with missing lines	Coverage Δ
src/spatialdata_plot/pl/render.py	89.01% <100.00%> (+0.02%)	⬆️

🚀 New features to boost your workflow:

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