Interface compatibility: Accept AbstractMatrix in getstructure

[ChrisRackauckas-Claude]

Summary

This PR improves interface compliance by accepting AbstractMatrix instead of just Matrix in the getstructure function.

• Changes getstructure(A::Matrix) to getstructure(A::AbstractMatrix) to support more array wrapper types (Symmetric, Adjoint, Transpose, SubArray views, etc.)

Interface Testing Performed

I tested the package against Julia's standard interfaces and SciML's array/number interfaces:

BigFloat Support ✓

All core functions work correctly with BigFloat:

• check_diagonal, is_toeplitz, is_banded, compute_bandedness - all pass
• getstructure - passes
• sparsestructure - works for Toeplitz, Banded, LowerTriangular, UpperTriangular
  • Note: isposdef fails for BigFloat sparse matrices, but this is a Julia SparseArrays limitation, not this package's issue

JLArray (GPU-like) Support

The package uses scalar indexing which is fundamental to the algorithms (element-by-element comparison for Toeplitz detection, banded detection, etc.). This is a design constraint rather than a bug - these algorithms inherently need to compare individual matrix elements.

ArrayInterface.jl Compatibility ✓

Works correctly with arrays that support fast_scalar_indexing.

Test plan

• All existing tests pass
• Verified getstructure now works with Adjoint, Symmetric, Transpose, and SubArray types
• Verified backward compatibility with regular Matrix inputs

cc @ChrisRackauckas

🤖 Generated with Claude Code

[ChrisRackauckas-Claude]

All tests pass locally with this interface improvement. The change from Matrix to AbstractMatrix in getstructure is a safe widening of the type signature that improves compatibility with wrapper types.

Ready to merge. cc @ChrisRackauckas

🤖 Generated with Claude Code