- Build lower triangular sparsity pattern (row >= col) instead of full
  symmetric matrix, so decompress_csc! with :L works correctly
- Always add diagonal entries explicitly (old Coloring module did this
  implicitly; SMC requires them to be in the sparsity pattern)
- Get CSC structure directly from the sparse matrix instead of going
  through compress/decompress
- Switch decompress_csc! from :U to :L to match lower triangular storage
- Update test expectations to CSC lower triangular order:
  J=[1,1,2] (was [1,2,1]), V=[3.4,2.1,1.3] (was [3.4,1.3,2.1])

https://claude.ai/code/session_01WBu9hZukriWDSSybN9gfBq

The SMC-based Hessian recovery stores nonzeros in lower triangular CSC
format (column-major: all entries of column j before column j+1, with
diagonal first in each column). Update all multivariate Hessian structure
and value expectations to match this ordering instead of the old
diagonal-first ordering.

https://claude.ai/code/session_01WBu9hZukriWDSSybN9gfBq

…ad of Base.promote_op

Base.promote_op may return Any for SMC.coloring since it lacks an explicit return
type annotation, causing TypeError when constructing _FunctionStorage{Any} because
Any <: AbstractColoringResult is false. Use a small test coloring call instead.

https://claude.ai/code/session_01WBu9hZukriWDSSybN9gfBq

… coloring

Pass a full symmetric sparsity pattern to SMC.coloring instead of lower
triangular, so the tree edges are correctly oriented. Store full_colptr,
lower_pos, and full_buffer in ColoringResult to allow recovery via
decompress_csc! with :F, then extract lower-triangular values.

This fixes: BoundsError: attempt to access 0-element Vector{Int64} at index [1]
in test_coloring_end_to_end_hessian_coloring_and_recovery.

https://claude.ai/code/session_01WBu9hZukriWDSSybN9gfBq