fix: support datetime variables in Dataset.interp

[emmanuel-ferdman]

PR Summary

Dataset.interp() silently dropped datetime64 and timedelta64 variables. Now they are interpolated by converting to float64 and back, with NaT handled like NaN.

• Closes Dataset.interp() silently dropping time-like data arrays #10900
• Tests added

[spencerkclark]

Thanks @emmanuel-ferdman—this largely looks good.

Could you add "Mm" to the valid dtypes on the DataArray.interp side as well? There is no need now to raise in that circumstance, which is great.

One edge case to consider is what to do in the scenario that extrapolation leads to float values outside the range that can be represented by 64-bit integers. Maybe we punt on that for now though, since it could be messy to handle in a robust way.

[spencerkclark]

I believe the following, by sticking completely to functions / methods defined for xarray objects, should be robust to all chunked array implementations, not just dask (computation comes from from xarray.computation import computation):

Suggested change

	var_indexers = {k: v for k, v in use_indexers.items() if k in var.dims}
	int_data = var.data.view(np.int64)
	nat = np.iinfo(np.int64).min
	as_float = np.where(
	int_data == nat, np.nan, int_data.astype(np.float64)
	)
	result = missing.interp(
	var.copy(data=as_float), var_indexers, method, **kwargs
	)
	as_int = np.where(
	np.isnan(result.data),
	nat,
	np.round(np.nan_to_num(result.data)).astype(np.int64),
	)
	variables[name] = result.copy(data=as_int.view(var.dtype))
	var_indexers = {k: v for k, v in use_indexers.items() if k in var.dims}
	int_data = var.astype(np.int64)
	nat = np.iinfo(np.int64).min
	as_float = computation.where(
	int_data == nat, np.nan, int_data.astype(np.float64)
	)
	result = missing.interp(
	as_float, var_indexers, method, **kwargs
	)
	as_int = computation.where(
	result.isnull(),
	nat,
	result.fillna(0).round().astype(np.int64)
	)
	variables[name] = as_int.astype(var.dtype)

[spencerkclark]

It was passing with your previous implementation, but this is always good to include as a sanity check for dask tests:

Suggested change

	result = ds.interp(x=[0.5, 1.5], y=[0.5, 1.5])
	with raise_if_dask_computes():
	result = ds.interp(x=[0.5, 1.5], y=[0.5, 1.5])