Fix stale comments

- hook_guard.hpp top block: described removal as `hook = {}` (destroy/reset); the
  model is now persistent disable_for_removal (never destroyed mid-session, the
  trampoline stays alive). Updated to match.
- input_source.hpp: SteamInputSource is no longer "future" -- it exists and is
  opt-in; reworded.
- audio_ring.hpp: format_generation actually bumps on every set_format (not
  "reserved, v1 sets once"); verify_capture is a 4-byte atomic guarded by the
  version gate (not "repurposed from a reserved byte old builds saw"); and the
  SharedBlock is no longer "20-byte pads".
- audio_format_verifier.cpp: dropped a dead `(void)recover_layout;` with a stale
  "step (a) only" comment -- the parameter is actually used.

Comment-only except the dead (void) cast.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

common/include/coop/audio_ring.hpp

@@ -2,9 +2,9 @@
 //
 // The injected hook (coop_hook.dll) captures the game's WASAPI render frames and
 // is the sole *producer*; the host (coop_host.exe) is the sole *consumer* and
-// re-renders the frames for Steam Remote Play Together. This is a separate,
-// larger mapping from the input/status SharedBlock (which is only 20-byte pads
-// and can't hold PCM): a header followed by a byte ring of `capacity` bytes.
+// re-renders the frames for Steam Remote Play Together. This is a separate, larger mapping from the
+// input/status SharedBlock (which holds fixed-size POD state, not bulk PCM): a header followed by a
+// byte ring of `capacity` bytes.
 //
 // Lock-free SPSC with free-running 64-bit positions (release on publish, acquire
 // on read) — the same cross-process atomic model as the input seqlock. POD and
@@ -58,9 +58,9 @@ struct AudioRingHeader
 	// (stream counting in HookStatus still runs regardless of this flag).
 	std::atomic<std::uint32_t> capture_enabled;
 
-	// Producer publishes the captured stream's format once, then sets
-	// format_valid=1 (release). format_generation is reserved so a future
-	// mid-session device re-init can be made forward-compatible; v1 sets once.
+	// Producer publishes the captured stream's format, then sets format_valid=1 (release).
+	// format_generation bumps on every (re)publish (audio_ring_set_format), so the host can notice a
+	// mid-session format change (a device re-init, or a measured-rate / override update).
 	std::atomic<std::uint32_t> format_valid;
 	std::atomic<std::uint32_t> format_generation;
 
@@ -92,8 +92,8 @@ struct AudioRingHeader
 	// capture both the hook (pre-mix) and a parallel process-loopback (post-mix) of the same audio
 	// and cross-correlate them to recover the true sample rate (and, in step b, channels/bit-depth)
 	// from ground truth instead of guessing. Inert (0) by default -- normal capture is unaffected,
-	// so it never changes the shipping no-echo path. Repurposed from `reserved`, so the layout and
-	// size are unchanged (old builds saw it as a zero reserved byte).
+	// so it never changes the shipping no-echo path. It's a 4-byte atomic carved out of the header's
+	// reserved space; the version gate (kAudioRingVersion) rejects any layout that doesn't match.
 	std::atomic<std::uint32_t> verify_capture;
 
 	std::uint8_t reserved[36];