Updated code to call o1

[tusharkhot]

@PeterAJansen Updated the README and code to call o1 models. The code works but it will need a lot more tokens to produce an output.

[PeterAJansen]

@MarcCote Tushar was mentioning that even with 10k output tokens, o1-mini didn't seem to be giving a response. It's possible it's spending all these tokens "thinking" (that's hidden from us), and not able to output anything. So running o1 might not be straightforward, and/or may require some prompt rewriting.

[MarcCote]

Ok interesting. I'll give it a try to check if at least the code is running on my end.

[tusharkhot]

Key thing to note @MarcCote

• Update recoma to 0.0.4
• Load the generator_params for o1 here: https://github.com/allenai/discoveryworld/blob/cc913cabc6b906e6dd0606fd3543159bc3c450e4/agents/recoma/configs/default_o1_generator.libsonnet