Add profile-based local run flows for Python samples

[zamderax]

Summary

• migrate python/web-app/wendy.json to profile-based configuration
• migrate python/yolov8/wendy.json to profile-based configuration with local + device targets
• ensure local YOLOv8 profile uses a writable model path (MODELS_DIR=../models) for host runs
• update YOLOv8 sample app/readme/runtime dependencies for the local profile workflow

Validation

• interactive wendy-dev run shows Local (this machine) row in the destination table
• local profile launch no longer fails on read-only /models path