fix(simulation): address PR #89 review — version comment, settings contract, docs, dead process recovery

Must-fix: protocol version comment now references PROTOCOL_VERSION
(no hardcoded number), settings delete is idempotent no-op.

Suggestions addressed: BehaviorModifier dedup claim dropped, hash
collision safety documented, FK pragma in test store, columns_to_value
consolidated, modifier_hint mismatch logging, OffDuty test coverage,
batching tradeoff documented, unknown value_type warning, occluded
text risk documented, dead child respawn in voice worker, INJECT
block ambiguity documented.

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

server/src/settings/mod.rs

@@ -139,6 +139,9 @@ pub fn process_settings_commands(
     };
 
     // Process all commands, last response wins (same tick batching).
+    // If multiple settings commands arrive in one tick, only the final response
+    // reaches the client. This is acceptable: the client sends one action per tick
+    // in normal use, and batch changes via RequestAll after bulk updates.
     for cmd in commands {
         match cmd {
             SettingsCommand::Change { ref key, ref value } => {
@@ -171,16 +174,13 @@ pub fn process_settings_commands(
                 });
             }
             SettingsCommand::Delete { ref key } => {
-                let deleted = store.delete(key);
+                // Idempotent: deleting a non-existent key is a silent no-op.
+                let _deleted = store.delete(key);
                 snapshot_buffer.pending_settings_response = Some(SettingsResponseWire {
                     success: true,
                     kind: "ack".into(),
                     settings: vec![],
-                    error: if deleted {
-                        None
-                    } else {
-                        Some(format!("setting not found: {}", key))
-                    },
+                    error: None,
                 });
             }
         }