Reconcile API docs and examples with the real library surface

[dchud]

Follow-up to #238 (bd-nj22). On the merged PR, @acdha noted the query-DSL sections still hand-rolled logic under headings named after the dedicated query types. Acting on that, I swept the recently-updated docs and everything nearby and found the same class of problem in several more places, plus reference tables and Python fences that did not match the compiled library.

Everything here was verified against the real surface: each corrected Rust snippet was compile-checked with a throwaway example, and each Python snippet was run against the built extension. The sweep also surfaced agent/intuition false positives that I confirmed are actually correct and left alone (e.g. ProducerConsumerPipeline.from_file exists; record.leader() is correctly a method; fields_by_tag does exist in Python; rust-api.md's Query DSL has_subfield is real).

Rust examples

• querying-fields.md: the Subfield Pattern/Value sections now use SubfieldPatternQuery + fields_matching_pattern and SubfieldValueQuery::new/::partial + fields_matching_value, instead of raw regex and manual string comparison (the exact thing @acdha flagged).
• reading-records.md: subfields_by_code; real MarcError struct variants (IoError { cause, .. }, InvalidLeader { message, .. }) — the old IoError(e)/InvalidRecord(msg) did not compile.
• concurrency.md: RecordBoundaryScanner + parse_batch_parallel(&boundaries, &buffer) (the old call used an undefined split_records and the wrong arity).
• encoding.md: record.leader.character_coding (there is no position_9()).
• testing.md: MarcReader instead of the nonexistent Record::from_marc21.

Reference tables and specialized records (rust-api.md)

• The Record and Field "Key Methods" tables now use real names (get_field/get_control_field/get_subfield), distinguish public fields (tag/indicator1/indicator2/leader/subfields) from methods, and show accurate return types (Option<&str>, iterators).
• The AuthorityRecord/HoldingsRecord examples use the real ::builder(leader) API (the old AuthorityRecordBuilder::new().control_number(...) did not exist).

Python fences

• get_fields for pymarc-style field["a"] access; record.leader() is a method; removed the nonexistent record.isbns() and field.ind1/ind2; leader.record_type.
• writing-records.md: modifying a field now uses remove + re-add, because in-place field edits do not persist to the record.

Encoding

• Removed documentation for MARC-8 output, which is not supported in either binding — MRRC writes UTF-8. There was no prior decision recording UTF-8-only-on-write; the encoder exists but is unwired.

Follow-up beads (code-level findings)

• bd-cdey — wire MARC-8 encoding into the writer, or commit to UTF-8-only and remove the dead encoder.
• bd-gmax — Python Record.fields_by_tag returns unwrapped (non-subscriptable) fields, unlike get_fields.
• bd-blja — Python in-place field edits are silently not persisted to the record.

.cargo/check.sh is green (566 tests + doctests + mkdocs).

Bead: bd-du5n

[codspeed-hq]

Merging this PR will degrade performance by 13.92%

⚠️ 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.

❌ 1 (👁 1) regressed benchmark
✅ 59 untouched benchmarks
⏩ 18 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
👁	WallTime	test_file_parallel_4x_10k_with_extraction	1.1 s	1.3 s	-13.92%

---

_{Comparing docs/query-tutorial-use-dsl (7d9e5af) with main (901af63)}

Footnotes

1. 18 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. ↩