Add CLI option to select TTS output target (robot or pc)

[hello-jkoubi]

Description

This PR introduces a new CLI option --tts_target to control where text-to-speech (TTS) audio is played when running stretch.app.ai_pickup.

Previously, all speech output was routed to the robot’s onboard speaker via the ROS2 bridge. This change adds flexibility by allowing users to instead play TTS audio locally on the workstation (e.g., GPU machine or connected speakers).

Usage

# Default (robot speaker)
python -m stretch.app.ai_pickup --use_llm

# Workstation speaker
python -m stretch.app.ai_pickup --use_llm --tts_target pc

[hello-robot-shehab]

Looks good Jason! Not sure how to to let it through though, the GitHub Actions pipeline is a bit broken atm, will look into it soon!