ENH: Add private RNG normalizer

[bruAristimunha]

Closes #9233

[larsoner]

I think we need to support legacy RandomState because scikit-learn still supports it (and actually does not support Generator IIRC), can you double check?

[larsoner]

Okay hopefully one last thing 🤞 For signatures like this, let's move the * to before the random-state. Some people might need to update their scripts to pass random_state= but probably not that many since this is so far down the list. And we've been trying to add * to callables anyway in reasonable places. Once that's done, random_state can be grouped with rng (and ideally placed immediately after it).

[larsoner]

@bruAristimunha one last-last idea... we've talked about how our random_state int values being 0 and 42 are not great. We should ideally randomly set these, at least in our examples and tutorials. Could you pick random values for those? The tricky part is that some of them might no longer work as expected, but in theory looking at the CircleCI output before and after to make a) things still execute without warnings and b) still look okay should make sure things are okay. You could also run examples locally but this might take a while...

[bruAristimunha]

doing @larsoner :)

[larsoner]

Minimal CI is unhappy, you need some exception for sklearn because it's not installed there

=================================== FAILURES ===================================
______________________________ test_no_global_rng ______________________________
[gw2] linux -- Python 3.12.14 /opt/hostedtoolcache/Python/3.12.14/x64/bin/python
mne/tests/test_docstring_parameters.py:379: in test_no_global_rng
    callables = _sklearn_callables(tree)
                ^^^^^^^^^^^^^^^^^^^^^^^^
mne/tests/test_docstring_parameters.py:340: in _sklearn_callables
    module = importlib.import_module(node.module)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
/opt/hostedtoolcache/Python/3.12.14/x64/lib/python3.12/importlib/__init__.py:90: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
<frozen importlib._bootstrap>:1387: in _gcd_import
    ???
<frozen importlib._bootstrap>:1360: in _find_and_load
    ???
<frozen importlib._bootstrap>:1310: in _find_and_load_unlocked
    ???
<frozen importlib._bootstrap>:488: in _call_with_frames_removed
    ???
<frozen importlib._bootstrap>:1387: in _gcd_import
    ???
<frozen importlib._bootstrap>:1360: in _find_and_load
    ???
<frozen importlib._bootstrap>:1324: in _find_and_load_unlocked
    ???
E   ModuleNotFoundError: No module named 'sklearn'
- generated xml file: /home/runner/work/mne-python/mne-python/junit-results.xml -

[larsoner]

(I'll cancel the others to save cycles)

[bruAristimunha]

Local seed-grid check for the simulated time-frequency example (0, 42, 181, and 271). The intended 50 Hz burst and smoothing trade-off remain visible with the new seed (181).

[larsoner]

Just had to fix a tiny bug with reST rendering, marking for merge-when-green, thanks in advance @bruAristimunha !