Move gym examples NN parameters in npy/npz/json files

[WilliamTambellini]

As today, the enjoy_TF_XXXXXX.py examples in
examples/pybullet/gym/pybullet_envs/examples
https://github.com/bulletphysics/bullet3/tree/master/examples/pybullet/gym/pybullet_envs/examples
are embedding the NeuralNet parameters directly in the python sources:
example:

bullet3/examples/pybullet/gym/pybullet_envs/examples/enjoy_TF_AntBulletEnv_v0_2017may.py

Line 70 in 04e3da3

weights_dense1_w = np.array(

Numpy has some serialization support:
https://numpy.org/doc/stable/reference/generated/numpy.lib.format.html#module-numpy.lib.format
It's even possible to store multiple arrays into a single npz file (a zip file containing npy arrays).

Would there be any cons to move these NN numpy arrays outside the python sources into some npy/npz (or even json) files ?
kind