Block internal Qt classes that should not be projected

[rcalixte]

These are some types that I've gone through and removed downstream where they aren't meant to be projected. This is why they have had stability issues that required patching. If they had been removed prior, we would not have needed to patch for 6.10.

[mappu]

I think I agree with this, I double-checked most of the classes and there's no public way of ending up with these types. They should generally be safe to remove from the bindings.

Notwithstanding, I think in future we will need bindings for some semi-private APIs for #302 and for QMetaObjectBuilder for run-time moc registration (#238 et al).

I think we've managed to avoid any breaking changes since the last tagged release, minor or otherwise, which I'm somewhat proud of, so i think i'd like to first tag a v0.13.0 and then merge this afterwards.

[rcalixte]

Notwithstanding, I think in future we will need bindings for some semi-private APIs for #302 and for QMetaObjectBuilder for run-time moc registration (#238 et al).

I didn't have any regressions for #302 downstream so I'm less worried about that but for #238, I don't have any QML exposure so I'm not actually sure if it would have an impact there. Between Qt 5 and QML, I'm trying to be careful not to introduce regressions from the changes that can be upstreamed since I'm only focused on Qt 6 for now. This repository has solid CI in place so I think I lean on that maybe more than I should. I'll run some spot checks to make sure nothing jumps out in the mean time and let you know (or adjust the PR) if anything comes up.