Fix regrid! for n-dimensional arrays and mapreduce arg order

[ctessum-claude]

Summary

• Fixes regrid! does not accept AbstractMatrix despite docstring claiming n-dimensional support #67: Add an AbstractArray method to regrid! that iterates over slices of the first dimension, enabling Matrix and higher-dimensional array inputs as promised by the docstring. The existing AbstractVector/DenseVector methods remain unchanged and take dispatch priority for vector inputs.
• Fixes mapreduce arguments swapped in cell_range_extent for ExplicitPolygonGrid{Planar} #65: Swap mapreduce argument order in cell_range_extent for ExplicitPolygonGrid{<: GO.Planar} from mapreduce(Extents.union, GI.extent, ...) to mapreduce(GI.extent, Extents.union, ...) (the mapping function should come first, the reducing function second).
• Add regression tests for vector, matrix, and 3D array regridding.

Test plan

• Existing vector regridding behavior unchanged (regression test included)
• Matrix (2D) regridding works correctly with uniform field
• 3D array regridding works correctly with uniform field
• Full test suite passes (1242 pass, 5 pre-existing broken)

🤖 Generated with Claude Code

[asinghvi17]

Hmm. I see what you mean here. Would you mind also documenting this so users don't trip over it? We may also need to break it at some point, or perhaps even allow (and require, a la eachslice) the user to specify which dim they want to slice over.