voice: quieter thinking tone, log cleanup, and small ui polish

- Add scripts/lower-thinking-tone.mjs to scale the thinking-tone PCM
  amplitude in-place; apply it at 0.1 so the cue is a soft indicator
  instead of triggering the VAD via mic feedback.
- Drop the "Received first voice_audio_chunk" log that re-fires every
  summary window because the chunk counter is reset.
- Match Spoke message icon and label sizing to the standard tool-call
  badge.
- Append a spoken-input instruction so voice replies route through the
  speak tool.

packages/server/src/server/session.ts

@@ -7546,15 +7546,6 @@ export class Session {
     const chunkBytes = Buffer.byteLength(msg.audio, "base64");
     this.voiceInputChunkCount += 1;
     this.voiceInputBytes += chunkBytes;
-    if (this.voiceInputChunkCount === 1) {
-      this.sessionLogger.info(
-        {
-          format: chunkFormat,
-          audioBytes: chunkBytes,
-        },
-        "Received first voice_audio_chunk for active voice mode",
-      );
-    }
     const now = Date.now();
     if (this.voiceInputChunkCount % 50 === 0 || now - this.voiceInputWindowStartedAt >= 1000) {
       this.sessionLogger.info(

packages/server/src/server/voice-config.ts

@@ -54,7 +54,7 @@ export function buildVoiceModeSystemPrompt(existing: string | undefined, enabled
 }
 
 export function wrapSpokenInput(text: string): string {
-  return `<spoken-input>\n${text}\n</spoken-input>`;
+  return `<spoken-input>\n${text}\n</spoken-input>\n<instruction>This message was spoken by the user. Respond using the speak tool only, not normal messages, because the user may not be looking at the chat.</instruction>`;
 }
 
 export function buildVoiceAgentMcpServerConfig(params: {