GeezerVison

First of four stacked PRs that complete \p{...} support with the properties backed by the Unicode Character Database, matched in C through a unicodedata capsule. This PR adds the machinery and the simplest tier: the enumerated properties stored as a single byte in the per-character record — Bidi_Class (bc), East_Asian_Width (ea), Grapheme_Cluster_Break (gcb) and Indic_Conjunct_Break (incb).

unicodedata exports the _ucd_re_CAPI capsule (the \N{...} precedent) and _ucd_re_info(), which lists the property selectors and value names. The parser resolves a property name and value to a selector and value index; the new CATEGORY_UCD opcode packs (negate, property, value) into one operand and is matched in C by sre_category_ucd(), which compares the index returned by the capsule. A negated single value is one charset item, so \P{bc=AL} composes inside a set.

The follow-up PRs add, in order: the numeric and binary properties (ccc, Bidi_Mirrored, Extended_Pictographic), the computed properties (Block, Decomposition_Type, Numeric_Type) and the remaining General_Category values and groups (Ll, Lo, M/P/S) with POSIX punct.

• Issue: re should support \p{...} character properties #95555