Add performance instrumentation and metrics

[Lucasmind]

Priority: LOW

Problem

There are no performance metrics, making it impossible to identify bottlenecks or optimize latency. Key measurements like time-to-first-token, TTS synthesis time, and end-to-end latency are not tracked.

Suggested Metrics

Metric	Where	Why
llm.time_to_first_token	LLMClient.js	Measures LLM responsiveness
llm.total_tokens	LLMClient.js	Tracks context usage
llm.tokens_per_second	LLMClient.js	GPU inference performance
tts.synthesis_time	TTSManager.js	TTS bottleneck identification
tts.chars_per_second	TTSManager.js	TTS throughput
tts.queue_depth	TTSManager.js	Queue saturation
stt.transcription_time	STTManager.js	STT latency
pipeline.end_to_end	Orchestrator.js	Total user-perceived latency
pipeline.time_to_speech	Orchestrator.js	Time from input to first audio

Suggested Implementation

// utils/metrics.js
class Metrics {
  constructor() {
    this.timers = new Map();
  }
  
  start(name) {
    this.timers.set(name, performance.now());
  }
  
  end(name, metadata = {}) {
    const start = this.timers.get(name);
    if (!start) return;
    const duration = performance.now() - start;
    this.timers.delete(name);
    log('info', `metric.${name}`, { duration_ms: Math.round(duration), ...metadata });
    return duration;
  }
}

export const metrics = new Metrics();

Files Affected

• New file: frontend/src/utils/metrics.js
• frontend/src/pipeline/Orchestrator.js
• frontend/src/llm/LLMClient.js
• frontend/src/tts/TTSManager.js
• frontend/src/stt/STTManager.js