Fix map() example in datasets documentation: define tokenizer before use

[Sanjaykumar030]

Problem

The current datasets.Dataset.map() example in the documentation demonstrates batched processing using a tokenizer object without defining or importing it. This causes a NameError when users copy and run the example as-is, breaking the expected seamless experience.

Correction

This PR fixes the issue by explicitly importing and initializing the tokenizer using the Transformers library (AutoTokenizer.from_pretrained("bert-base-uncased")), making the example self-contained and runnable without errors.
This will help new users understand the workflow and apply the method correctly.

Closes #7703

[Sanjaykumar030]

Hi @lhoestq, just a gentle follow-up on this doc fix PR (#7704). Let me know if any changes are needed — happy to update.
Hope this improvement helps users run the example without confusion!