Reuse query-string separators and start fields directly
Evaluate the query-string proposal from @deadstrobe5 in issue #365. Reuse the & position found while scanning a field name, and start the next field directly when its first byte is already available. Duplicate and trailing separators keep the existing path.
Add regression tests for field-start events across every two-chunk split and byte-by-byte writes, and for max_size ending at a separator. Both tests also pass against unchanged main.
Local wall-clock timings on CPython 3.13.18, macOS arm64, for the existing large query-string input: 41.3 to 30.7 µs with no-op callbacks, and 54.7 to 43.1 µs with append callbacks. These are local results, not a confirmed portable speedup; the existing CodSpeed benchmark will evaluate the change in CI. No multipart or workflow changes are included.
Validation: scripts/check, scripts/test --strict-config --strict-markers (187 tests, 100% coverage), and scripts/rename pass on Python 3.13.18. An additional 39,360 differential cases against main matched callbacks, consumed bytes, and errors across chunking, strict parsing, and size limits.
Session: pi --session 01a1e0a2-78a9-7473-86ce-b992bd92cf74