BUG: misleading documentation in scipy.stats.entropy

Reproducing Code Example

The entropy function allows calculation of entropy along an axis for multi-variate distributions. It works as expected, and returns an ndarray as a result. However in the documentation, the return value is specified as a float (which is indeed returned for univariate distributions).

Thus the problem is in the documentation, and not the implementation itself. Additionally, as a consequence the IDE flags this as a type mismatch when type hinting is used as it assumes the function really return a float and not an array.

To reproduce:

from scipy.stats import entropy
import numpy as np
pk = np.array([[2/3, 1/3],[1/3, 2/3]])
e = entropy(pk)

### SciPy/NumPy/Python version information

`1.7.1 1.21.2 sys.version_info(major=3, minor=9, micro=6, releaselevel='final', serial=0)`

[YairMZ]

@tupui Thanks for the quick response. I’d love to help and make a PR. The truth is however, that my coding experience has been mostly on my my own projects, and therefore I don’t really know what exactly do I need to do. If you could direct me, I’d be happy to help.

[tirthasheshpatel]

Closed in #14685