Add Features and Improvements for Lunar Lake Release

[mschuettlerTNG]

Description:

This PR adds several new features and improvements in preparation of the upcoming LNL release.

Changes Made:

• New Features
  • The Conversation History in the Answer tab allows to save multiple conversation threads across AI-Playground sessions.
  • Additional LLMs have been added and can be selected in the Answer tab. As some of these might be gated (i.e. they require a huggingface.co account with a granted access request to the model), there is now the possibility to add a huggingface.co token via the AI-Playground settings
  • An Aspect Ratio Picker has been added to AI-Playground settings to make it easier to select useful and valid resolution for image generation in the Create tab.
• Improvements
  • Improved Error Handling during Startup should make the AI-Playground more reliable and provide better feedback to the user in case of issues during startup.
  • A Dynamic Port Selection during the python backend initialization should prevent certain startup issues from occurring.
  • There is now a clear indication when an image generation has been blocked by the safety checker, instead of just displaying a black picture.
  • The font size of the conversation in the Answer tab can now be adjusted.
  • There is now a dismissible performance warning when switching to HD mode.

Testing Done:

The changes have been tested on a Intel® Arc™ A750 system.

Checklist:

• I have tested the changes locally.
• I have self-reviewed the code changes.
• I have updated the documentation, if necessary.