Refactor classification example to avoid hardcoded label token IDs

[Ashish-Patnaik]

This PR improves the robustness of "examples/classification.py" by removing hardcoded
token IDs for the classification labels.

Fixes Issue #496

Changes

1. Dynamically computes the token IDs for "Yes" and "No" using the tokenizer instead
   of relying on fixed numeric IDs.
2. Adds a small validation check to ensure each label maps to exactly one token,
   preventing silent errors if tokenization changes.
3. Fixes minor typos in the prompt template and comments:
   • "grammaticaly" = "grammatically"
   • "respectivelly" = "respectively"

Why?
Hardcoding token IDs makes the example fragile if the tokenizer vocabulary changes
or if the script is reused with a different model variant. This change keeps the
example functionally identical while making it safer and easier to maintain.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.