BUG: support DataFrame columns containing period arrays

[sb123sb123]

Summary

ipmt fails when its per input is a pandas Series of same-shaped period arrays and the other inputs are row-wise Series. Stack the period arrays into a row-by-period matrix, expand matching row parameters into columns, and broadcast them together. Apply the same normalization in ppmt and add regression coverage for both functions without adding a pandas test dependency.

Fixes #105.

Validation

• pytest -q: 111 passed (Python 3.14, NumPy 2.5.3, pandas 3.0.6).
• spin lint: Ruff, Pyright, and Mypy passed.
• The issue reproduction passed on Python 3.11.4, NumPy 1.26.4, and pandas 2.2.3; the TestIpmt/TestPpmt group passed (38 tests). The source-only checkout lacked the compiled _cfinancial extension, so a temporary import stub was used for that focused test run and removed afterward.

AI assistance was used to prepare this change and its tests.

[mattip]

Is there any point in adding the typing info when it is all Any?

[sb123sb123]

You were right: the six NDArray[Any] entries did not narrow the helper’s dynamic broadcast result. I removed the return annotation and kept the _ArrayLike input annotations. Ruff, Python 3.13.15 py_compile, and the focused nested-period regression test pass (1 passed). The source checkout lacks the compiled _cfinancial extension, so the focused test used a temporary import stub; this test path does not call that extension.

[sb123sb123]

The revised head is 59c55de. The focused regression, Ruff, and Python 3.13 syntax checks pass locally. GitHub has created Test package and Type-check runs for this commit, but both are currently action_required and have not executed. Could a maintainer approve those two runs when convenient? I will address any resulting failures.