BUG: stats.ttest_1samp: fix use of keepdims

[sdiebolt]

Reference issue

Partially addresses gh-20725.

What does this implement/fix?

This PR fixes an error raised when using the keepdims argument of scipy.stats.ttest_1samp. The error comes from the _axis_nan_policy_factory decorator that tries to expand the dimensions of all returns from decorator functions. Some of the decorated functions return simple integers that cannot be expanded, thus the error:

scipy/scipy/stats/_axis_nan_policy.py

Lines 240 to 246 in 626bbd1

	def _add_reduced_axes(res, reduced_axes, keepdims):
	"""
	Add reduced axes back to all the arrays in the result object
	if keepdims = True.
	"""
	return ([np.expand_dims(output, reduced_axes) for output in res]
	if keepdims else res)

Additional information

test_keepdims originally raised an error when adding stats.ttest_1samp due to a division-by-zero warning. The error is due to some test cases where omitting NaNs leads to samples of size 1, thus the division-by-zero when computing the SEM. To fix this, I set the too_small argument to a callable that checks if the sample to test has at least size 2. I could not simply use too_small=1 however, since it seems popmean is considered a sample (ie., n_samples is set to 2 in the decorator of ttest_1samp).

Adding all other functions from axis_nan_policy_cases raises many more errors that I'll try to check now.

[sdiebolt]

@mdhaber Ok so it seems most errors that occur when running test_keepdims on all axis_nan_policy_cases are due to small sample sizes (occurring for a number of reasons, from samples becoming too small after omitting NaNs, to samples being simply too small or having a degenerate dimension, e.g. with the case sample_shape=(20, 0)). What would you recommend? Should I try to modify test_keepdims parameters in this PR to fit axis_nan_policy_cases?

[mdhaber]

In that case, maybe leave the existing test alone if it's that strict. Instead, add a new test like it that doesn't generate such cases, since they seem orthogonal to the issue of non-expandable scalars. The inconsistent behavior w.r.t. small samples is being fixed separately; when that's resolved, we can probably combine the tests more easily.

[mdhaber]

This fixes a bug as-is, so I'd like this to make it into 1.14.0.
We just did a major cleanup of warnings/errors for functions with sample sizes <= too_small. We need to do a follow-up of functions that need a different too_small than they have, so I don't want that to hold up or be touched in this issue.
I'll leave gh-20725 open after this so we remember to add other functions to test_keepdims after fixing too_small.

[mdhaber]

If tests pass, I'll merge.
@tylerjereddy can we backport this fix into the next release candidate?