Add Advanced Options to Control Gemini Safety and Harm Probability

[daverad]

We often get errors back where the error is "Safety" when using Gemini. We believe adding a parameter to adjust the safety sensitivity may help us get meaningful responses. The questions and content we are sending are safe - so its unclear why gemini is flagging to begin with, but ability to edit these settings would be helpful.

https://ai.google.dev/api/generate-content#harmprobability