Add __repr__ methods for DatasetColumn and DatasetColumnPair

[zanvari]

Summary

This PR addresses #547 by adding __repr__() implementations for DatasetColumn and DatasetColumnPair.

Currently, both classes inherit from list but do not populate the underlying list. As a result, printing either object displays [], even though they provide lazy access to dataset columns. This can make debugging difficult and lead to confusing exception messages and logs.

The new representations provide a concise summary of each object without loading dataset contents into memory. For DatasetColumn, the representation includes the column key and dataset length. For DatasetColumnPair, it includes the configured column names, key mappings, and dataset length.

Changes

• Added __repr__() to DatasetColumn
• Added __repr__() to DatasetColumnPair
• Added tests covering both representations

Example

Before:

print(DatasetColumn(...))
# []

After:

print(DatasetColumn(...))
# DatasetColumn(key='text', len=3)

Similarly, DatasetColumnPair now provides a meaningful representation instead of displaying as an empty list.

Testing

I added tests for both classes and verified that the new representations are returned as expected.

python -m pytest tests/test_evaluator_utils.py -v

Result:

2 passed

Fixes #547.