in operator with dtype and set does not work as expected

When testing if the dtype of an array is present in a set of dtypes, the answer is unexpected.

Reproducing code example:

import numpy as np
f32_dtype = np.zeros(1, dtype=np.float32).dtype

# test with list of dtypes
print(f32_dtype in [np.float32, np.uint8])  # => True (as expected)

# test with set of dtypes
print(f32_dtype in {np.float32, np.uint8})  # => False (not as expected)

Error message:

none

Numpy/Python version information:

numpy 1.18.4 Python 3.7.4 (tags/v3.7.4:e09359112e, Jul 8 2019, 20:34:20) [MSC v.1916 64 bit (AMD64)]

[eric-wieser]

Agreed that this is a bit unpleasant - we might be able to adjust dtype.__hash__ to fix this.

[seberg]

Going to close this as duplicate of gh-7242.