Refactor audio handling

[ryze312]

Hey, here is another PR from me! Made a few changes to audio handling and cleaned up some of the code. In particular audio devices get started/stopped on demand upon entering voice call. This allows audio backends like PipeWire to suspend sinks when they there isn't anything active connected to them.

AudioManager:

• Audio devices get started/stopped on voice connect/disconnect
• Reduce code duplication with Open/TryOpen functions
• Don't store device configs and ids

AudioDevices:

• Rename [Get/Set]ActivePlaybackDevice to [Get/Set]ActivePlaybackDeviceIter
• Declare getters as const and [[nodiscard]]
• Reduce code duplication by using Get[Playback/Capture]DeviceIDFromModel
• Add GetActive[Playback/Capture]

TODO:

• [x] Start/stop voice audio devices on demand
• [x] Move notification sounds handling to SystemAudio
• [x] Start/stop notification audio device after timeout as well
• [x] Split voice audio into multiple devices per user
• [x] Add options for persistent audio devices and split voice audio