Qiao Zhang 65ef487a82 Allow jnp.nan_to_num handle integer types like numpy.
See current behavior difference wrt np.nan_to_num
```
>>> np.nan_to_num(np.array(1, dtype=np.int32))
1
>>> jnp.nan_to_num(jnp.array(1, dtype=jnp.int32))
ValueError: data type <class 'numpy.int32'> not inexact
```
PiperOrigin-RevId: 505735212
2023-01-30 10:37:17 -08:00
..
2023-01-30 00:24:30 -08:00
2023-01-12 22:58:31 -08:00
2021-01-13 10:26:35 -05:00