Speed up datashader rendering of points

[timtreis]

Summary

Closes #379

Datashader was consistently slower than matplotlib for rendering points due to five performance bottlenecks in the datashader code path. This PR fixes all five:

1. Dask DataFrame passed to cvs.points() — PointsModel.parse() returns dask, but datashader's dask path has ~137x scheduler overhead on already-computed data. Fix: .compute() before aggregation.
2. Double extent computation — extent was computed via get_extent() on dask (triggering compute), then .compute() was called again. Fix: new _datashader_canvas_from_dataframe() that reads min/max directly from the materialized pandas DataFrame.
3. Per-point _hex_no_alpha() calls — O(n) list comprehension even when no alpha stripping was needed. Fix: deduplicate + skip when all colors already in #RRGGBB format.
4. _build_datashader_color_key iterated all points — used expensive per-element pandas Index lookups and couldn't early-exit because sentinel categories inflated the target count. Fix: numpy arrays + count only present categories.
5. _want_decorations O(n) set creation — set(cv.tolist()) on 500K+ hex strings. Fix: numpy (cv == cv.flat[0]).all().

Performance results (after JIT warmup)

No coloring:

n	matplotlib	datashader (before)	datashader (after)	speedup
10K	0.065s	0.997s (0.07x)	0.051s	1.22x
100K	0.145s	0.267s (0.58x)	0.108s	1.32x
500K	0.540s	0.763s (0.75x)	0.383s	1.37x
1M	1.040s	0.861s (1.21x)	0.724s	1.43x

Categorical coloring:

n	matplotlib	datashader (after)	speedup
100K	0.168s	0.154s	1.09x
500K	0.605s	0.381s	1.59x

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 73.84%. Comparing base (edca5a5) to head (e1f1ad4).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #557      +/-   ##
==========================================
- Coverage   73.89%   73.84%   -0.06%     
==========================================
  Files          10       10              
  Lines        2777     2783       +6     
  Branches      645      644       -1     
==========================================
+ Hits         2052     2055       +3     
- Misses        451      454       +3     
  Partials      274      274              

Files with missing lines	Coverage Δ
src/spatialdata_plot/pl/_datashader.py	89.05% <100.00%> (-2.87%)	⬇️
src/spatialdata_plot/pl/render.py	82.10% <100.00%> (-0.04%)	⬇️
src/spatialdata_plot/pl/utils.py	65.67% <100.00%> (+0.19%)	⬆️

🚀 New features to boost your workflow:

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