Add get_max_output_tokens for class SpyrePlatform

[gmarinho2]

FIX: #148

Currently the class SpyrePlatform uses the default max_tokens set in the OpenAI frontend code. This new method selects the warmup shape that fits the prompt and has the biggest shape['new_tokens']. This way SpyrePlatform will use the maximum number possible for token generation when max_tokens is not set in the request body.

[github-actions]

👋 Hi! Thank you for contributing to vLLM support on Spyre.
Just a reminder: Make sure that your code passes all the linting checks, otherwise your PR won't be able to be merged. To do so, first install the linting requirements, then run format.sh and commit the changes. This can be done with uv directly:

uv sync --frozen --group lint --active --inexact

Or this can be done with pip:

uv pip compile --group lint > requirements-lint.txt
pip install -r requirements-lint.txt
bash format.sh

Now you are good to go 🚀

[wallashss]

I think you should probably set this PR as draft and wait for the feedback of the vllm community on vllm-project/vllm#18557. This changes depends on that.

[maxdebayser]

LGTM, once the upstream PR gets merged.