optimise hot-path audio pipeline: zero-copy ring buffer, pre-allocated converter buffers

- Replace Swift Array<UInt8> ring buffer with UnsafeMutableRawPointer to
  eliminate COW ref-count checks on every write/read
- Add append(from:count:) to copy directly from Core Audio buffer pointer
  into the ring buffer, removing the per-callback Data heap allocation
- Pre-allocate AVAudioPCMBuffer pair in AudioFormatConverter and reuse
  across transform() calls (lazy init, capacity-checked)
- Fix float-to-int truncation in output frame count calculation (ceil)
- Add comprehensive AudioBuffer test suite (12 tests) including proper
  wrap-around coverage for both append and read paths

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Sources/AudioTeeCLI/AudioTee.swift

@@ -45,7 +45,8 @@ struct AudioTee {
       name: "exclude-processes", help: "Process IDs to exclude (space-separated)")
     parser.addFlag(name: "mute", help: "Mute processes being tapped")
     parser.addFlag(name: "stereo", help: "Records in stereo")
-    parser.addFlag(name: "flush", help: "Flush stdout after each audio chunk (reduces latency when piping)")
+    parser.addFlag(
+      name: "flush", help: "Flush stdout after each audio chunk (reduces latency when piping)")
     parser.addOption(
       name: "sample-rate",
       help: "Target sample rate (8000, 16000, 22050, 24000, 32000, 44100, 48000)")