runtime/audio (macOS): voices on one AVAudioEngine instead of an AVAudioPlayer per sound

AVFAudio keeps a 64-byte AudioQueueOwner for good on every AVAudioPlayer play
after the clip has finished - measured one a play, whatever is called around
it (prepareToPlay, pause, no rewind) and still there after the player is
released; a stop before the play made one every time. The windowed walk showed
it as AudioQueueOwner 73 -> 87 in a minute.

A sound is now decoded once into a PCM buffer and played by a voice of its own
on one shared engine: a player node (the buffer scheduled again on each play,
looping for -1), a varispeed (the rate) and a small mixer (volume and pan).
snd_playing compares the uptime clock with the end worked out when the clip was
played (asking the node where it is made two AVAudioTime objects a call), and
snd_play drains an autorelease pool of its own. The C interface is unchanged.

A harness driving snd_* directly (400 plays past the end, the playing flag
checked during and after each, a loop and a stop, the setters) holds the heap
flat to a block and gets the flag right 400 of 400; the windowed valley
compiles and links against it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

packages/ludic.audio/load.ludic

@@ -1,5 +1,5 @@
 # load.ludic - a clip the runtime can actually open. Audio.load takes a real FILESYSTEM path (it
-# bottoms out in AVAudioPlayer), not the pack-aware file_open every other asset arrives through: in a
+# bottoms out in AVAudioFile), not the pack-aware file_open every other asset arrives through: in a
 # bundle every clip loads as 0 while Fs.exists, which IS pack-aware, says it is there. So try the
 # runtime's path, and on 0 lift the bytes out of the pack into the cache directory and load that.