Updates for model specific prompting to better show citations

[dayland]

In this PR we have added new logic to:

• Lookup the AOAI deployment to retrieve the underlying model and version
• Updated the prompt engineering to be specific to model type as we have found that trying to use one system prompt for both gpt-35-turbo and gpt-4 led to degradation of one or the other.
• Added new management Service Principal that is used to query the AOAI deployment via Azure Cognitive Services Management python library
• Added extra values to the "Info" panel in the UX to show the model name and version and other useful information
• Updated the App Service configuration values to use conditional check to properly populate when not using an existing AOAI instance
• Changed max token paramater value to 1024 for generated responses when making API call to generate final response to avoid token overflow error.

Fixes #162
Fixes #174

[ArpitaisAn0maly]

Reviewed. Approved. Updated comment section to reflect token count change to API call.