Reginald production workflow

[rchan26]

Aim is to have two models running:

• llama-index-llama-cpp running on a VM accessible via API call and host slackbot on Azure running with all_data index
• llama-index-azure-gpt running on Azure with public index

Tasks:

• Set up pulumi to host slack bot (Update pulumi script(s) #113)
  • Set API_URL environment variable
• Set up VM to turn on and off at set times
  • Write a startup script that VM runs on startup to start listening for calls
• Write in some logic in slack bot for how to respond if the VM is off (using try/excepts)
• Update current hackweek Slack bot to use updated "public" indices?
• Build indices on Azure container that save to a file share which the VM can access