Use of np.float and np.int etc

[psteinb]

Thanks @Gabri95 et al for providing this library. I however discovered a small bug, that makes e2cnn unusable with any numpy version higher or equal to 1.24.

The numpy team deprecated the np.float and np.int and np.bool aliases to the respective python internals.
numpy.org/doc/stable/release/1.20.0-notes.html#using-the-aliases-of-builtin-types-like-np-int-is-deprecated
From numpy 1.24 and higher, their use produces an error which makes the example given in the README.md break immediately.

This is a doublicate of QUVA-Lab/escnn#25 but for this library.