Fix contours() docstring overstating Dask peak-memory guarantee

[Melissari1997]

Summary

Closes #2787. The contours() docstring in xrspatial/contour.py tells Dask users:

For Dask inputs, each chunk is processed independently and results are merged, keeping peak memory proportional to chunk size.

The implementation does not deliver that guarantee. The global merge step (_deduplicate_by_level) materialises all contour segments at once to stitch polylines across chunk boundaries, so peak memory scales with total contour complexity, not chunk size. Cross-chunk stitching inherently needs all segments for a level in memory simultaneously.

Change

xrspatial/contour.py — corrected the Notes section to accurately describe the memory trade-off.

Testing

No behavioural changes — the fix edits only a docstring.

[Melissari1997]

PR Review: Fix contours() docstring overstating Dask peak-memory guarantee (#2787)

Suggestions (should fix, not blocking)

None — the single change is correct and narrowly scoped.

What looks good

• The new wording is accurate: chunking bounds per-chunk scan buffers, but the global merge (_deduplicate_by_level at contour.py:469) materialises all segments at once. The old phrasing ("keeping peak memory proportional to chunk size") was misleading and is correctly removed.
• The diff is minimal (4 lines changed) and touches only the docstring — zero behavioural risk.
• Existing tests pass unchanged, confirming no logic was touched.

Checklist

• Algorithm matches reference/paper — N/A (docstring only)
• All implemented backends produce consistent results — N/A
• NaN handling is correct — N/A
• Edge cases are covered by tests — N/A
• Dask chunk boundaries handled correctly — N/A
• No premature materialization or unnecessary copies — N/A
• Benchmark exists or is not needed — N/A
• README feature matrix updated — N/A
• Docstrings present and accurate — ✓