Consider retaining sex in `anonymize_info()`

[cbrnr]

Currently, anonymize_info() preserves age (instead of the full birthday), which is both anonymized and useful for many analyses (and there is no way to disable this behavior). The same argument seems to apply to sex, but this field is currently dropped (set to 0 for unknown).

In #8130, there appeared to be consensus that keeping sex would be acceptable, but it was never implemented (or exposed via a parameter). Would it make sense to retain sex by default (in line with keeping age), or at least add a parameter to control this? I'm happy to submit a PR.

[cbrnr]

I just noticed that keep_his=True actually retains not only his_id, but also sex and hand (see https://github.com/mne-tools/mne-python/blob/main/mne/_fiff/meas_info.py#L3721-L3731). This is undocumented behavior, so we definitely need to fix something here. I suggest that keep_his=True only retains his_id, and that we keep sex and hand (just like we keep age). Alternatively, I would add two new parameters keep_sex=False and keep_hand=False, but then we should discuss dropping the age as well in order to have a "fully anonymized" data set in the sense that not even age is kept.