Codebase simplification: fix delete_subfield, reduce duplication

[dchud]

Summary

Addresses the high and medium priority items from #81. Net reduction of 253 lines while fixing a bug and improving consistency.

High priority

• Fix delete_subfield() bug: Was returning the value without actually removing the subfield. Added Field::delete_subfield() in Rust core, exposed via PyO3, updated Python wrapper to delegate. Now matches pymarc's behavior exactly.
• Fix recovery_mode validation inconsistency: AuthorityMARCReader and HoldingsMARCReader now reject invalid recovery_mode values with ValueError, matching MARCReader behavior from Add to_unicode, permissive, and recovery_mode flags to MARCReader #80.
• Extract _wrap_field() / use _wrap_record() helpers: Replaced 16+ inconsistent field wrapping patterns and duplicated record wrapping in __init__.py with two shared helpers.
• Extract shared reader_helpers.rs: Moved source detection (path/bytes/file), recovery mode parsing, and Python file I/O into a shared module. Reduced authority + holdings readers from 766 → 541 lines (~225 lines eliminated).

Medium priority

• Extract control_field_char_at() utility: Replaced 5 duplicated length-check + chars().nth() patterns across authority and holdings record 008 field parsing.

Test plan

• All 775 Rust tests pass (lib + integration + doc)
• All 611 Python tests pass
• Full .cargo/check.sh passes (rustfmt, clippy, ruff, tests, doc tests)
• delete_subfield() manually verified to match pymarc behavior

Bead: bd-1dr9
Closes #81

🤖 Generated with Claude Code

[codspeed-hq]

Merging this PR will degrade performance by 20.11%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 10 improved benchmarks
❌ 7 (👁 7) regressed benchmarks
✅ 43 untouched benchmarks
⏩ 16 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
👁	WallTime	test_pipeline_parallel_4x_10k_threaded	92.3 ms	114.5 ms	-19.35%
👁	WallTime	test_pipeline_parallel_extraction_4x_10k_threaded	125.7 ms	140.2 ms	-10.36%
⚡	WallTime	test_threaded_reading_1k	28.3 ms	25.3 ms	+11.68%
⚡	WallTime	test_threaded_reading_4x_1k	94.2 ms	83.8 ms	+12.46%
👁	WallTime	test_pipeline_sequential_extraction_4x_10k	94.8 ms	110.1 ms	-13.9%
👁	WallTime	test_process_4_files_parallel_4_threads	93.4 ms	116.9 ms	-20.11%
⚡	WallTime	test_threaded_reading_4x_10k	991.8 ms	854.3 ms	+16.09%
👁	WallTime	test_process_8_files_parallel_4_threads	197.2 ms	245.9 ms	-19.81%
⚡	WallTime	test_parallel_read_with_extract_4x_1k	100.7 ms	90 ms	+11.86%
⚡	WallTime	test_threading_speedup_4x_10k	992.8 ms	848.5 ms	+17%
⚡	WallTime	test_file_parallel_4x_10k_with_extraction	1,085.2 ms	853 ms	+27.23%
👁	WallTime	test_pipeline_sequential_1x_10k	18.7 ms	21.3 ms	-12.04%
👁	WallTime	test_pipeline_sequential_4x_10k	78.2 ms	87.5 ms	-10.64%
⚡	WallTime	test_file_parallel_4x_10k	1,037.4 ms	825.8 ms	+25.62%
⚡	WallTime	test_parallel_read_4x_10k	983.8 ms	864.2 ms	+13.84%
⚡	WallTime	test_parallel_read_with_extract_4x_10k	1,072.6 ms	908.2 ms	+18.1%
⚡	WallTime	test_threaded_with_title_extraction_4x_10k	1,008 ms	864.6 ms	+16.59%

---

_{Comparing bd-1dr9-codebase-simplification (e808d71) with main (ae3a684)}

Footnotes

1. 16 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩