Fix some potential memory leak in image capture sheet

• Use camerax to implement image capture in-app instead of using the native camera app to reduce the possibility of app being killed due to GPU OOM.
• Show accelerator name in chat message sender labels.
• Attach download workers to silent foreground notifications to make them less likely to be killed.
• Update app icon to be consistent with Google style.

• Update Gemma 3n 4B model.
• Update context length to 4K for Gemma 3n models.
• Fix a bug related to LLM inference engine cleanup.
• Various UX improvements.

• Fix a download resume bug

• Check version when app resumes.

• Fix a bug where deleting a local model when there are >1 local models shows the wrong download status.
• Don't go back to model selection screen automatically when there is an error during model initialization, so that users have a chance to change model parameters (e.g. accelerator) to retry the initialization.
• Show error dialog properly in prompt lab screen.

• Better text formatting on home screen for smaller screen.
• Set default backend based on the order of "accelerators" field in model allowlist.

• Re-enable image picker menu items when model config is updated.
• Make theme picker work when system is in dark mode.

• Save the loaded model allowlist to a local file so that it can be read when the allowlist cannot be loaded from internet.
• Improve the image clipping transitioning from full image back to the image thumbnail in chat ui.