Implement NumPy's `__array_function__` protocol for array methods that are not in the Array API Standard

[tomwhite]

This is to support nanmean, nansum (see #153), and pad (#461).

The previous attempt at this in #468 was reverted in #469, since it was causing problems when used with Xarray (https://github.com/pydata/xarray/issues/8834#issuecomment-2124914727).

When Xarray is working we can re-introduce the changes here.