Audio Spatializer Improvements

[bjornbytes]

Summary of changes:

• Enables Steam Audio by default in all LÖVR builds!
• Upgrades Steam Audio from version 2 to version 4.8 (Upgrade phonon to v4 #474)
• Add a new AudioMesh object, with dynamic transform (Add dynamic audio geometry support #866, lovr.audio.setGeometry with phonon is kinda slow #406)
• Audio meshes can have per-triangle audio materials (Per-triangle AudioMaterials #409).
• Adds lovr.audio.setHRTF to enable a custom HRTF. Binaural spatialization will only work if an HRTF has been set, otherwise the spatializer will fall back to simple panning.
• Add t.audio.debug to get log messages from Steam Audio and miniaudio.

TODO (maybe not all in this PR):

• Add lovr.audio.commit, use triple buffered audio updates (Atomic Audio Functions #892)
• Remove lock from audio thread
• Run reverb simulation asynchronously on a background thread
• Add support for effect tails (Phonon reverb can get cut off #404)
• Add support for stereo blend
• Add support for ambisonic sound playback with binaural spatialization and rotation
• Consider some way to encode ambisonic sounds
• Consider modifying attenuation/directivity APIs to be more flexible
• Add lovr.audio.capture to make it easier to get mic samples without using a sink