FEAT: streaming support in fetchone for nvarcharmax data type

[gargsaumya]

Work Item / Issue Reference

AB#38110
AB#34162

GitHub Issue: #<ISSUE_NUMBER>

---

Summary

This pull request improves NVARCHAR data handling in the SQL Server Python bindings and adds comprehensive tests for NVARCHAR(MAX) scenarios. The main changes include switching to streaming for large NVARCHAR values, optimizing direct fetch for smaller values, and adding tests for edge cases and boundaries to ensure correctness.

NVARCHAR data handling improvements:

• Updated the logic in ddbc_bindings.cpp to use streaming for large NVARCHAR/NCHAR columns (over 4000 characters or unknown size) and direct fetch for smaller values, optimizing performance and reliability.
• Refactored data conversion for NVARCHAR fetches, using std::wstring for conversion and simplifying platform-specific handling for both macOS/Linux and Windows.
• Improved handling of empty strings and NULLs for NVARCHAR columns, ensuring correct Python types are returned and logging is more descriptive.

Testing enhancements:

• Added new tests in test_004_cursor.py for NVARCHAR(MAX) covering short strings, boundary conditions (4000 chars), streaming (4100+ chars), large values (100,000 chars), empty strings, NULLs, and transaction rollback scenarios to verify correct behavior across all edge cases.

VARCHAR/CHAR fetch improvements:

• Improved direct fetch logic for small VARCHAR/CHAR columns and fixed string conversion to use the actual data length, preventing potential issues with null-termination and buffer size. [1] [2]

[sumitmsft]

Left a few comments

[bewithgaurav]

will wait for latest main branch merge, will refactor this PR to a great extent

[gargsaumya]

will wait for latest main branch merge, will refactor this PR to a great extent

The conflicts are now resolved. You can go ahead and re-review.