docs(bump): correct height_func contract and count default in docstring

[brendancol]

Fixes two accuracy problems in the bump() docstring, found during a documentation sweep of the module.

• height_func was described as taking x, y per point. bump() actually calls it once with a (count, 2) locations array and expects a length-count heights array back. The per-point form raises TypeError when used literally.
• count was typed int with no default. It is optional and defaults to min(width * height // 10, 10_000_000).

Docstring-only, no behavior change. The runnable example already used the array-based height_func convention, so it needed no edit.

Closes #3606

[brendancol]

PR Review: docs(bump): correct height_func contract and count default in docstring

Documentation-only change to bump() in xrspatial/bump.py. I verified the corrected claims by running the code against the worktree build.

Blockers (must fix before merge)

None.

Suggestions (should fix, not blocking)

None.

Nits (optional improvements)

• xrspatial/bump.py:184-185: the count default is written as min(width * height // 10, 10_000_000). When agg is passed, width/height are ignored and the dimensions come from agg.shape (lines 337-338, 344). The formula is still correct in terms of the raster size, but a reader who only passes agg might wonder where width/height come from. Optional: phrase it in terms of the output raster dimensions.

What looks good

• The old height_func description (function which takes x, y) was wrong: the code calls height_func(locs) once with a single (count, 2) int32 array (line 394), so a literal f(x, y) callable raises TypeError. I confirmed both the new contract and the TypeError on the old one.
• The new count doc matches count = min(w * h // 10, _MAX_DEFAULT_COUNT) with _MAX_DEFAULT_COUNT = 10_000_000 (lines 349-350, 29).
• The height_func=None default is documented as height 1, matching lambda bumps: np.ones(len(bumps)) (line 387).
• The docstring's own plot example already uses the (count, 2) contract (heights(locations, ...) reading locations.shape[0] / locations[r]), so example and prose are now consistent.

Checklist

• Docstring matches actual behavior (verified by execution)
• Parameter types and defaults documented
• [n/a] Algorithm/backend/NaN/dask correctness — no code changed
• [n/a] Benchmark / README feature matrix — docs-only, no API surface change

[brendancol]

PR Review (follow-up): docs(bump): correct height_func contract and count default

Re-review after merging origin/main and applying the one nit from the first pass.

What changed since the last review

• Merged origin/main (was 19 commits behind, PR showed CONFLICTING). The only conflict was in the bump() docstring, where main's #3610 had rewritten the count line and added the name= parameter while still carrying the old, wrong height_func text. Resolved by keeping this PR's corrected height_func contract and main's name= parameter (signature name: str = 'bump', doc entry, and name=name on both DataArray returns are all present). PR is now MERGEABLE.
• Applied the nit I raised: the count default now reads min(w * h // 10, 10_000_000), where w/h are the output raster dimensions (taken from agg when given), instead of naming width/height which are ignored on the agg path.

Blockers / Suggestions

None.

Nits

None outstanding.

Verification

• pytest xrspatial/tests/test_bump.py — 20 passed against the worktree build.
• Post-merge diff vs base is confined to the bump() docstring and the sweep state CSV. No code paths changed.

Checklist

• Merge conflict resolved preserving both main's name= and this PR's corrections
• Docstring matches actual behavior (verified by execution earlier)
• Tests pass post-merge
• [n/a] Algorithm/backend/NaN/dask correctness — no code changed