Implement additional gemini backend

[andreashappe]

see https://ai.google.dev/gemini-api/docs/models/gemini

[lloydchang]

workaround: use a proxy such as https://github.com/zhu327/gemini-openai-proxy

[lloydchang]

Regarding https://developers.googleblog.com/en/gemini-is-now-accessible-from-the-openai-library/

Content safety settings are not supported via Gemini's OpenAI chat completion API, as shown in the example at https://www.reddit.com/r/GoogleGeminiAI/comments/1ffb4by/is_there_a_way_to_change_the_safety_settings/

[lloydchang]

Consider a workaround, such as a proxy, that turns off the content safety settings by default.

For example:
gemini-openai-proxy at https://github.com/zhu327/gemini-openai-proxy/blob/559085101f0ce5e8c98a94fb75fefd6c7a63d26d/pkg/adapter/chat.go#L230-L246
and
litellm at https://github.com/BerriAI/litellm/blob/70aa85af1fc01d30934ee71a703b8d3982420d05/litellm/llms/vertex_ai_and_google_ai_studio/vertex_ai_non_gemini.py#L203-L213

The above nuances are better suited for projects that act as proxies, such as gemini-openai-proxy, litellm, etc.

Adding a Gemini backend to hackingBuddyGPT would require ongoing maintenance to adapt to continuously evolving changes in Gemini's API, which is in beta (v1beta) rather than OpenAI's non-beta (v1) API.

For instance, maintainers and contributors might spend time trying to understand and justify why Gemini v1beta offers an OpenAI chat completion API but does not fully support content safety settings:

• https://www.reddit.com/r/GoogleGeminiAI/comments/1ffb4by/is_there_a_way_to_change_the_safety_settings/
• https://www.reddit.com/r/LocalLLaMA/comments/1gn3zhh/gemini_is_now_accessible_from_the_openai_library/

In conclusion, using proxies is a more sensible approach, given the current state of the Gemini v1beta API, as it helps avoid the overhead of adapting to its ongoing changes.

[lloydchang]

Relates to Mac target localhost container via gemini openai proxy #94
create onboarding/quickstart blog post as well as video #36
docs(README.md): add Mac use case #95