hdbscan: add support to other types of input integers

[jcfaracco]

Both attributes like min_samples and min_cluster_size can be defined
using other types of integer like Numpy's int16, int32 and int64. This
should be also supported by HDBSCAN. This commit only extendes the check
to other types of integers.

For further example, you can test using:

import numpy as np

a = 10
b = np.int16(10)
c = np.int64(10)

print(isinstance(a, int))
print(isinstance(b, int))
print(isinstance(c, int))
print(np.issubdtype(type(a), np.integer))
print(np.issubdtype(type(b), np.integer))
print(np.issubdtype(type(c), np.integer))

The output is:

True
False
False
True
True
True

Signed-off-by: Julio Faracco jcfaracco@gmail.com

[lmcinnes]

Thanks. Hopefully no one needs to use a full size int64 for min_samples, but not erroring if it happens to be typed that way is very useful.