Fix multiple pipecat implementation issues and runtime errors

This commit provides a comprehensive set of fixes to address a series of
`ModuleNotFoundError`, `ImportError`, `TypeError`, and `RuntimeError`
exceptions that were preventing the `pipecatapp` from running.

The changes include:

- **API Updates:** The application code has been updated to use the latest
  APIs for `pipecat-ai`, `piper-tts`, and `RealtimeSTT`. This resolves
  multiple `ImportError` and `TypeError` exceptions.
- **File Deployment:** The Ansible playbook has been updated to copy all
  necessary Python modules and static assets to the application's runtime
  environment, which resolves the `ModuleNotFoundError` exceptions.
- **Model Caching:** The Ansible playbook now pre-caches the `faster-whisper`
  and `silero-vad` models, which resolves network-related errors that
  were occurring in the server environment.
- **Path Correction:** The `web_server.py` script has been updated to use
  dynamic paths for locating static assets, making the application more
  portable.
- **Dependency Management:** The `webrtcvad` dependency has been added to
  `requirements.txt`.
- **File Creation:** All missing files (`memory.py`, `web_server.py`,
  `moondream_detector.py`, `static/index.html`, and the `tools`
  directory and its scripts) have been created.
- **Code Restoration:** The original functionality of the tool scripts and
  the `memory.py` script has been restored.

Author: google-labs-jules[bot]

ansible/roles/pipecatapp/files/app.py

@@ -109,8 +109,6 @@ def reset(self):
         self.llm_first_token_time = 0
         self.tts_first_audio_time = 0
 
-from piper.voice import Piper
-
 # (Your existing WebSocketLogHandler, UILogger, and BenchmarkCollector classes remain here)
 # ...