Pass empty values to Csv post-processors for None input - #188

Csv normally passes an iterator of cast values to post_process, but its None branch calls the same function without an argument. As a result, Csv(cast=int, post_process=sum)(None) raises TypeError while the equivalent empty-string input returns 0. A custom post-processor with a required argument also fails when used with Config(..., default=None).

Normalize None to an empty string and reuse the existing parsing path. This supplies an empty iterator to the post-processor without calling the item cast, and preserves the empty list/tuple results.

The regressions cover None and empty-string input with list, tuple, sum, and sorted, plus a custom iterator-consuming post-processor through Config(default=None). Before the fix, three regressions failed and nine CSV tests passed.

Validation:

• Python 3.10.21: all 76 tests passed.
• Python 3.12.3: all 76 tests passed; 97% coverage.
• PyPy 2.7.18 with the project's pinned pytest/mock versions: 74 passed, including all CSV regressions. Two existing mock_open encoding tests failed identically in an untouched baseline checkout.
• Source/test byte compilation and git diff --check: passed.
• Wheel/sdist build and twine check: passed, with the existing warning about missing long_description_content_type.

This issue was identified during source review. The implementation and tests were prepared with OpenAI Codex assistance.