Add negation flag to SubfieldPatternQuery

[dchud]

Motivation

A user scanning ~250k records needed to find ones where a subfield did NOT match a regex. Currently this requires Python-side filtering (not re.match(...)) which means:

• Extra string copies across the Rust/Python FFI boundary for every record
• GIL overhead on each field value that crosses into Python
• Inability to short-circuit inside Rust iteration

Adding a negate flag pushes the inversion into Rust so non-matching fields never cross the FFI boundary.

Design

Follow the existing partial flag pattern on SubfieldValueQuery — add a negate: bool field defaulting to false.

Rust core (src/field_query.rs)

• Add pub negate: bool to SubfieldPatternQuery
• Update new() to set negate: false
• Add pub fn negated(tag, subfield_code, pattern) -> Result<Self> constructor (sets negate: true)
• Update matches(): use != self.negate so that when negated, a match means "subfield exists but does NOT match the pattern"

PyO3 bindings (src-python/src/query.rs)

• Add negate as an optional keyword argument to PySubfieldPatternQuery::new() (default False)
• Add #[getter] fn negate(&self) -> bool
• Update __repr__ to include negate=true when set

Python stubs (mrrc/_mrrc.pyi)

• Update SubfieldPatternQuery.__init__ signature: add negate: bool = False
• Add negate property

Tests

• Rust unit tests in src/field_query.rs: negated match, negated non-match, negated with missing subfield (should return false — subfield must exist)
• Python tests in tests/python/test_query_dsl.py: negated pattern query via kwarg, verify negate property, integration with fields_matching_pattern()

Documentation

• Update docs/guides/query-dsl.md with negation examples
• Update docs/tutorials/python/querying-fields.md if SubfieldPatternQuery is covered there
• Update API reference docs if they document SubfieldPatternQuery parameters

Edge cases

• Missing subfield: When negate=true and the subfield doesn't exist on a field, matches() returns false. Negation means "the subfield exists but its value doesn't match" — not "the subfield is absent."
• Repeating subfields: get_subfield() returns only the first subfield with the given code. With negate=true, only the first value is checked. This is consistent with existing non-negated behavior but should be documented.

matches() implementation

field.get_subfield(self.subfield_code)
    .is_some_and(|value| self.pattern.is_match(value) != self.negate)

Scope

This issue covers the negate flag on SubfieldPatternQuery only. SubfieldValueQuery negation is tracked separately. A broader query composition framework (NotQuery, AndQuery, OrQuery) is a separate concern.

Bead: bd-w9cm