Add Reasoning Effort Parameters for o3 and o1 models

[faev999]

I think so far we can choose o3-mini-high directly, but it would be nicer if we could set the reasoning effort for models like O3 and O1