[FEATURE] - Add rpm_limit argument

[patrickfleith]

Is your feature request related to a problem? Please describe.

Gemini models offer free generations but are capped in terms of RPM, RPD.

• LLM providers or the generate method should offer to deliberately slow down the generation process to accomodate low RPM.
• This could be down via an rpm_limit argument.

Imagine two providers: OpenAI + Gemini:

• OpenAI is not rate limited
• Gemini we want to leverage the firee tier so it is rate limited: say 10 RPM.
  We should set rpm_limit=10 either in GeminiProvider class or 'LLMProvideror ingenerate()`. I have the feeling it should be at provider level. TBC.

[patrickfleith]

Closed with #95