Add support for greater token lengths

[danaimone]

Currently if you try to use the gpt-4o model with >2048 token length, the following exception occurs upon prompting:

thread '<unnamed>' panicked at ata/src/prompt.rs:81:20:
called `Option::unwrap()` on a `None` value
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

I am requesting that we add support for a great token length, as the pricing and performance is much better.

Reproduction:

• Change model to gpt-4o, token length >2048
• Try prompting