minor update for supporting 3rd party Open AI APIs + add Ollama

[cyyeh]

minor modification after merging from community's contribution to support 3rd party Open AI APIs: https://github.com/Canner/WrenAI/pull/365

• add OPENAI_API_BASE to .env files
• add embedding models as env
• add Ollama provider and changed it to async version
• allow launcher to use custom llm
• related new docs PR: https://github.com/Canner/docs.getwren.ai/pull/13

steps to test custom llm:

• install ollama and download these models
  • ollama pull llama3:8b
  • ollama pull nomic-embed-text
• follow the guide(Running Wren AI with your Custom LLM or Document Store) here: https://github.com/Canner/docs.getwren.ai/blob/e1345364c10395791e1bc995a41cc0307255501b/docs/installation/custom_llm.md
• fill in ollama to LLM_PROVIDER in .env.ai, and these are default models used for Ollama, so you don't need to setup GENERATION_MODEL, EMBEDDING_MODEL, EMBEDDING_MODEL_DIMENSION. However, if you would like to try other LLMs, then you need to fill in these environment variables