Fix isort import-ordering drift in bump.py

[brendancol]

Closes #3609

What

Run isort on xrspatial/bump.py to fix a Cat 4 import-ordering finding. The from xrspatial.utils import (...) block was laid out one name per line; the configured isort (line_length=100) wraps it as a two-line grid instead.

Categories addressed

• Cat 4 (isort import ordering). No behavior change intended.
• Cat 1/2/3/5 were clean, so nothing else in scope: no flake8 E/W codes (baseline 0), no unused imports or dead locals, no bare except, no mutable default args, no shadowed builtins.

Test plan

• flake8 xrspatial/bump.py clean
• isort --check-only xrspatial/bump.py clean
• xrspatial/tests/test_bump.py passes (18 tests, CUDA available so cupy and dask+cupy paths ran)
• Module imports and all five xrspatial.utils names resolve

Found by the /sweep-style deep sweep (2026-07-02).

[brendancol]

PR Review: Fix isort import-ordering drift in bump.py

This is a lint-only change. The one code edit reflows the from xrspatial.utils import (...) block from a one-name-per-line layout to the two-line grid wrap that the configured isort (line_length=100) produces. The imported names are unchanged (ArrayTypeFunctionMapping, _validate_scalar, has_cuda_and_cupy, is_cupy_array, ngjit), so there is no runtime effect. The other diff hunk is the style-sweep state CSV row.

Blockers (must fix before merge)

None.

Suggestions (should fix, not blocking)

None.

Nits (optional improvements)

None. The reformat matches isort output exactly and both resulting lines (91 and 50 columns) stay under the 100-column limit, so flake8 is still clean.

What looks good

• No behavior change: same names, same order, only the wrapping differs.
• flake8 baseline was already 0 and stays 0; isort is clean after the change.
• The 18-test bump suite still passes with CUDA available, so the cupy and dask+cupy paths were exercised.

Checklist

• Algorithm matches reference/paper (unchanged; no algorithm touched)
• All implemented backends produce consistent results (import-only change; tests pass on all four)
• NaN handling is correct (unchanged)
• Edge cases are covered by tests (existing coverage unaffected)
• Dask chunk boundaries handled correctly (unchanged)
• No premature materialization or unnecessary copies (no logic change)
• Benchmark exists or is not needed (not needed for a lint fix)
• README feature matrix updated (not applicable)
• Docstrings present and accurate (unchanged)