Do not require `rng` in `verify_grad`

[ricardoV94]

Description

Just create one on-demand np.random.default_rng(). No need to require it explicitly?

import pytensor.tensor as pt
from pytensor.gradient import verify_grad

verify_grad(pt.log, [2.0])
# TypeError: rng should be a valid instance of numpy.random.RandomState. You may want to use tests.unittest_tools.verify_grad instead of pytensor.gradient.verify_grad.

The message is also outdated, as numpy.random.Generator are accepted (and preferred)