Generate arguments for virtual functions that take float as taking float (rather than double)

[dsnopek]

While looking into issue #1350, I noticed that we were generating virtual methods that were declared to take float in Godot as taking double in godot-cpp.

For example, AudioStreamPlayback::_mix(AudioFrame *, float, int) would turn up in godot-cpp as AudioStreamPlayback::_mix(AudioFrame *, double, int). This is despite the extension_api.json giving float as the "meta".

So, this PR makes it so that those functions actually take float. This fits with our general goal of providing the same API on both the Godot and godot-cpp side.

I don't know if there's any problems with doing this. However, it seems like it should be fine, because our PtrToArg<T>::convert() should be able to change the double * passed in the ptrcall into a float for the function call.

But I'm going to mark it as draft for now, while I investigate the implications of this change.

Note: This doesn't actually fix the problem I was looking into originally. It's entirely possible that this is totally unrelated to that issue. :-)

[fire]

float means double in gdscript. I am not sure the exact way to resolve this.

[AThousandShips]

The meta type matches the typing of the actual c++ method though, the same is done in C#

[dsnopek]

Yeah, when it's a float in Godot's C++ API, GDScript's double would be converted to a float at some point anyway.

Here it's a question of how we expose Godot's C++ API's that take floats to developers using godot-cpp. I feel like they should see a float if the API uses a float in Godot's C++, since we want to match the API between Godot and godot-cpp. And this could also help avoid the possible misconception that all the precision of the double would be maintained, since it would be converted to/from a float at some point anyway.

I just need to make sure that I'm implementing this the right way, and it isn't going to causes some other problem elsewhere.