Update jnp.fabs to emulate the behavior of np.fabs for complex inputs.

[rajasekharporeddy]

This PR fixes a discrepancy between jnp.fabs and np.fabs by raising a TypeError for complex inputs, as np.fabs currently does.

Current behavior:

>>> x = jnp.array([1-1j, 2+5j])
>>> jnp.fabs(x)
Array([1.4142135, 5.3851647], dtype=float32)

New behavior:

>>> x = jnp.array([1-1j, 2+5j])
>>> jnp.fabs(x)
TypeError: ufunc 'fabs' does not support complex dtypes

Also updated the docstring with code examples. The rendered preview:

• https://jax--23290.org.readthedocs.build/en/23290/_autosummary/jax.numpy.fabs.html

[jakevdp]

Looks good! Since this is a user-visible change of behavior, we should also add a changelog entry.