Propagate the units of the convolution kernel.

[roytsmart]

convolve() is annotated kernel: np.ndarray | u.Quantity, but only the
unit of array was ever unwrapped. A Quantity kernel reached the compiled
kernels still wrapped, where Numba treats it as a plain array, and the
outcome depended on whether array had a unit too:

import numpy as np, astropy.units as u, ndfilters

a = np.arange(7.0) * u.mm
k = (np.ones(3) / 3) * u.s

ndfilters.convolve(a, k).unit             # mm  -- the kernel unit is dropped
ndfilters.convolve(np.arange(7.0), k)     # AttributeError: 'numpy.ndarray'
                                          # object has no attribute 'unit'

The kernel is now unwrapped alongside the array and the two units are
multiplied together, so those give mm s and s respectively. A
convolution is a sum of products, so the product of the units is the unit of
the result.

Tests

Added a case over the four combinations of a unitless, dimensionless, and
dimensional array and kernel, asserting the exact unit of the result and
that the values match the unitless convolution. The existing suite covers a
Quantity array with a bare kernel.

Verified locally: pytest passes (2883 tests, up from 2875), pyright reports
0 errors, and black --check and ruff check ndfilters/ are clean.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (9ee208d) to head (2fb4750).

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #35   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           14        14           
  Lines          651       677   +26     
=========================================
+ Hits           651       677   +26     

Flag	Coverage Δ
unittests	100.00% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.