Add matrix support for AkimaInterpolation

[mrarat]

Summary

This PR adds AbstractMatrix input support for AkimaInterpolation.

Previously, AkimaInterpolation(u, t) only accepted vector-like u, while the documented interface supports stacked array data where the last dimension corresponds to time. With this change, matrix-valued data can be passed as:

u::AbstractMatrix  # rows are output components, columns correspond to t
t::AbstractVector
A = AkimaInterpolation(u, t)

For scalar evaluation times, A(t_eval) now returns a vector containing the interpolated value for each row of u.

Changes

• Added an AkimaInterpolation(u::AbstractMatrix, t; ...) constructor.
• Added matrix-valued Akima interpolation evaluation.
• Added matrix-valued Akima derivative support for in-domain evaluation.
• Added matrix-valued Akima integral support for in-domain integration.
• Added tests comparing matrix Akima behavior against independent row-wise scalar Akima interpolations.

Known limitation

This PR intentionally does not fix matrix-valued extrapolation.

The current generic extrapolation code in DataInterpolations uses first(A.u) and last(A.u) to obtain endpoint values. For
matrix-valued u, those return scalar array entries, not the first/last output column. This affects matrix-valued extrapolation more generally, not only Akima.

Because of that, matrix-valued extrapolation and extrapolated derivative/integral behavior should be addressed in a separate PR with a global fix.

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.

Additional information

• Fixes Add vector output support for AkimaInterpolation #468.
• This PR builds on the earlier work in Add missing matrix support for AkimaInterpolation and QuadraticSpline #467. I reviewed that implementation and kept the scope focused on matrix-valued AkimaInterpolation construction, in-domain evaluation, derivatives, and integrals.