v0.3

This project's been dormant for a while and the codebase seems stable enough, so I figure I should make a release.

Internally clownaudio now operates in S16 instead of F32. This should hopefully lead to performance improvements as sample-conversion between the decoder backends and the mixer will be bypassed. Likewise, S16 uses less memory and may even be faster depending on integer vs. float performance differences.

Additionally, miniaudio, stb_vorbis, dr_flac, dr_mp3, and dr_wav have been updated, and a buffer overflow in PxTone has been corrected.