Add examples for processing Q&A and instruction-following datasets in exercises

[leozanchett]

No description provided.

[leozanchett]

This PR enhances the dataset processing examples in the course.

Changes:

Added a clear separation between generic Q&A processing and instruction-following datasets.

Included synthetic dataset examples using datasets.Dataset.from_dict to make the code snippets self-contained and runnable without external dependencies.

Added a summary section to guide learners on when to use each specific processing function (process_qa_dataset, process_instruction_dataset, or custom ones like process_gsm8k).

Improved code readability with structured headers and print statements for better output visualization.

Reasoning:
The previous version only showed the GSM8K example. By adding explicit examples for the generic functions, students can better understand how to apply these tools to their own custom datasets immediately.