feat: add Google AI Gemini Pro support

[cpacker]

Example Gemini API + MemGPT usage

Setting up Gemini with MemGPT configure:

$ memgpt configure
Loading config from /Users/loaner/.memgpt/config
? Select LLM inference provider: google_ai
? Enter your Google AI (Gemini) API key (see https://aistudio.google.com/app/apikey): *********
? Enter your Google AI (Gemini) service endpoint (see https://ai.google.dev/api/rest): generativelanguage
? Select default model: gemini-pro
Got context window 30720 for model gemini-pro (from Google API)
? Select your model's context window (see https://cloud.google.com/vertex-ai/generative-ai/docs/learn/model-versioning#gemini-model-versions): 30720
? Select embedding provider: openai
? Select default preset: memgpt_chat
? Select default persona: sam_pov
? Select default human: basic
? Select storage backend for archival data: chroma
? Select chroma backend: persistent
? Select storage backend for recall data: sqlite
📖 Saving config to /Users/loaner/.memgpt/config

Testing an agent using the Gemini API:

$ memgpt run
Loading config from /Users/loaner/.memgpt/config

? Would you like to select an existing agent? No

🧬 Creating new agent...
->  🤖 Using persona profile: 'sam_pov'
->  🧑 Using human profile: 'basic'
Loading config from /Users/loaner/.memgpt/config
Loading config from /Users/loaner/.memgpt/config
🎉 Created new agent 'CapriciousDaffodil' (id=602e6e68-be69-480e-a744-374d329ccb9f)

Hit enter to begin (will request first MemGPT message)

💭 This is more like it! Hello there! This is indeed my first login. Let's see if I can access my 
memory core.
🤖 Hey there! Excited to be here.

---

Please describe the purpose of this pull request.

• Adds support for Google AI Gemini
• Refactors the llm_api_tools.py file into a nested directory for all code related to calling LLM API services (via REST)
  • "LLM API services code" = code where we call a server that supports ChatCompletion format with tool/function calling
• Refactors the HTTP request code to:
  • Use ChatCompletionRequest as the payload/data arg type (to match ChatCompletionResponse in the return)
  • Pass around List[Message] instead of List[dict] in the calls to the APIs (get_ai_reply)
    • This makes it easier to convert to arbitrary payload/request types (such as Gemini) that are slightly different from OAI format

How to test
How can we test your PR during review? What commands should we run? What outcomes should we expect?

Have you tested this PR?

• Test OpenAI
• Test Azure
• Test local
• Test Gemini
• Test summarize method

[sarahwooders]

lgtm!

[quantumalchemy]

Request upgrade to gemini 1.5 pro api
--> gemini-1.5-pro-latest
Add allow option / switch --> safety_settings=None