BUG/TST: Include sem & count in all_numeric_reductions

[mroeschke]

The sem bug with Arrow types could be backported to 1.5.x but given the larger testing changes IMO it's fine for 2.0

[jbrockmendel]

is kwargs['ddof'] not relevant here?

[mroeschke]

Yeah I suppose this only applies in the stddev call.

When comparing to ser.astype("Float64").sem() the results were off, and removing kwargs["ddof"] aligned this pyarrow sem to nullable dtype sem.

[jbrockmendel]

the test changes all look good. this change i dont know enough to form an opinion on. if youre confident its right then LGTM. otherwise lets find an appropriate person to double-check

[mroeschke]

I think based on the nanops.nansem implementation, I'm fairly sure these align

pandas/pandas/core/nanops.py

Lines 1033 to 1042 in 3fffb6d

	nanvar(values, axis=axis, skipna=skipna, ddof=ddof, mask=mask)
	mask = _maybe_get_mask(values, skipna, mask)
	if not is_float_dtype(values.dtype):
	values = values.astype("f8")
	count, _ = _get_counts_nanvar(values.shape, mask, axis, ddof, values.dtype)
	var = nanvar(values, axis=axis, skipna=skipna, ddof=ddof)
	return np.sqrt(var) / np.sqrt(count)

[mroeschke]

Going to push this one in since the pyarrow sem implementation matches the nanops.sem one. If both are wrong at least we have a test now that said both should be aligned.