Fix flake8 E201 in edge_detection kernel literals (#3675)

[brendancol]

Closes #3675

Style cleanup from a /sweep-style pass over xrspatial/edge_detection.py. Categories addressed: Cat 1 (E201, E302) and Cat 4 (isort, test file only).

• Drop the space after [ in the SOBEL_Y and PREWITT_Y kernel literals (the 4 E201 lines flake8 reports on this file). Matches the LAPLACIAN_KERNEL style already in the file; kernels are numerically identical.
• test_edge_detection.py: add the missing second blank line before the first fixture (E302) and apply isort's import-block reflow for the configured line_length=100.

No behavior change. The module only wraps convolve_2d, so backend coverage is untouched (numpy / cupy / dask+numpy / dask+cupy all go through convolution).

Test plan:

• flake8 xrspatial/edge_detection.py xrspatial/tests/test_edge_detection.py clean
• isort --check-only clean on both files
• 87 tests in test_edge_detection.py pass locally (CUDA host, GPU tests run)
• Kernel arrays verified equal to their previous values

[brendancol]

PR Review: Fix flake8 E201 in edge_detection kernel literals (#3675)

Blockers (must fix before merge)

None.

Suggestions (should fix, not blocking)

None.

Nits (optional improvements)

• xrspatial/edge_detection.py:12-14 and 21-23: dropping the leading space loses the first-column alignment that lined the 0/1 rows up under the -1 rows. A per-line # noqa: E201 would have kept it, but the new form matches LAPLACIAN_KERNEL a few lines down, so the file is at least internally consistent now. Fine as is.

What looks good

• The kernel edits are whitespace-only. SOBEL_Y and PREWITT_Y still equal their previous values, and none of the five public functions changed.
• The test-file import reflow is what the configured isort (line_length=100) produces, and the E302 fix takes the file to zero flake8 findings.
• No runtime code paths are touched. All four backends route through convolve_2d, which this PR leaves alone.
• State CSV row for edge_detection is one physical line and follows the existing schema.

Checklist

• Algorithm matches reference (kernels numerically unchanged)
• All implemented backends produce consistent results (no runtime change)
• NaN handling is correct (untouched)
• Edge cases are covered by tests (existing 87 tests pass)
• Dask chunk boundaries handled correctly (untouched)
• No premature materialization or unnecessary copies
• Benchmark exists or is not needed (style-only change)
• README feature matrix updated (not applicable, no new functions)
• Docstrings present and accurate (unchanged)