fix: validate max_new_tokens against speculative_tokens during warmup (fixes #474)

[mithilesh1909]

… (fixes #474)

What does this PR do?

When speculative decoding is enabled, generating up to the model's max positional embeddings causes a CUDA device-side assert error at runtime. This PR adds an early validation check in the router that rejects requests where max_new_tokens + speculative_tokens would exceed the available token budget (max_total_tokens - input_length), returning a clear 422 error instead of crashing with a CUDA error.

Changes:

• Added speculative_tokens field to the Validation struct
• Added MaxNewTokensSpeculative error variant with a descriptive message
• Added validation check in both tokenized and non-tokenized input paths
• Added a unit test covering the boundary conditions
• Updated Validation::new() call in server.rs to pass speculative_tokens

Fixes #474

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Was this discussed/approved via a Github issue or the discord / slack channel? Please add a link
  to it if that's the case. Improve warmup checking for max new tokens when using speculative decoding #474
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.