feat: Add Whisper GPU speech-to-text with progressive transcription

- Add faster-whisper Python server for GPU-accelerated transcription
- Support dual mode: Web Speech API or Whisper GPU (toggleable)
- Progressive transcription every 3 seconds while recording
- Separate terminal server process (stable during hot-reload)
- Add Ctrl+V paste and Ctrl+C copy support in FloatingTerminal
- Add MCP tools: whisper_start, whisper_stop, whisper_toggle, whisper_status
- Update package.json with separate api/terminal/frontend processes

server/terminal.ts

@@ -0,0 +1,22 @@
+#!/usr/bin/env bun
+/**
+ * Terminal Server - Independent process
+ * This runs separately from the main server to maintain stable Claude Code sessions
+ * even when the main server restarts due to code changes.
+ */
+
+import { startTerminalServer } from './services/terminal'
+import { WORKING_DIR } from './config'
+
+console.log('')
+console.log('='.repeat(50))
+console.log('Terminal Server (Independent Process)')
+console.log(`  WebSocket: ws://localhost:4103`)
+console.log(`  Working Dir: ${WORKING_DIR}`)
+console.log('')
+console.log('This process is stable and won\'t restart')
+console.log('when the main server reloads.')
+console.log('='.repeat(50))
+console.log('')
+
+startTerminalServer()