[macOS/iOS] Use hardware sampling rates for audio I/O.

[bruvzg]

Instead of "max_rate" from the project settings, use preferred OS values, since API can work only with the limited set of values depending on device.

Fixes #58180

TODO:

• I'm not sure if changes in the AudioStreamPlaybackMicrophone are sufficient to correctly resample it.
• We should track device changes and adjust sampling rates accordingly, but I'm not sure if Godot audio subsystem is capable of handling sampling rate changes.

[kus04e4ek]

I'm still new to the Godot's source code, so take everything I say with a grain of salt.

I'm not sure if changes in the AudioStreamPlaybackMicrophone are sufficient to correctly resample it.

They look like they are. AudioStreamPlaybackMicrophone is a child of AudioStreamPlaybackResampled, AudioStreamPlaybackResampled::mix calls _mix_internal and it uses get_stream_sampling_rate and AudioDriver::get_mix_rate to correctly resample audio.

I'm not sure if Godot audio subsystem is capable of handling sampling rate changes.

Since audio generation is happening only in a callback by the current output device and uses AudioDriver::get_mix_rate to resample all of the audio, as long as AudioDriver::get_mix_rate returns the correct mix rate for the current output device and AudioDriver::get_capture_mix_rate returns the correct mix rate for the input device that last wrote to the buffer everything should be fine (AudioDriverCoreAudio::output_callback, then AudioDriver::audio_server_process, then AudioServer::_driver_process, then AudioServer::_mix_step, then playback->stream_playback->mix or AudioStreamPlaybackResampled::mix)

[kus04e4ek]

get_input_mix_rate is probably a better name. See #69120