Remove nameparser.config.regexes in 3.0? The parser has not read REGEXES since 2.0 #1008

nameparser.config.regexes is public, but the parser reads none of it. Its 19 REGEXES entries configured the v1 parser through CONSTANTS.regexes. Since 2.0, parsing is configured by named Policy fields, and the docs already say both halves:

• docs/customize.rst: editing REGEXES "is not a supported override. The parser does not read REGEXES at all."
• docs/migrate.rst: every regexes.* entry except bidi and emoji "has no 3.0 replacement". Those two became Policy(strip_bidi=...) and Policy(strip_emoji=...).

So the module configures nothing and describes nothing about how a name is parsed. The vocabulary modules beside it raise a related but separate question, in #1007.

What keeping it costs

• A third copy of every pattern the pipeline uses. Layering keeps _pipeline/ and _render from importing nameparser.config, so they hold their own copies. tests/v2/test_regex_sync.py then pins those copies to config/regexes.py as the "source of truth". That is backwards: the pipeline's copy is the one that parses, and the config copy is the one that exists only to be compared against. Петров Иван Никитич reads family 'Никитич' under locales.RU: Никитич and Саввич are missing from the irregular patronymics #994 edited the East Slavic patronymic pattern in three files where two would have done: the pipeline, the config table, and the RU pack's DEVIATES copy, which test_locales.py already holds to the pipeline.
• A mutable dict that looks like configuration. Rename the vocabulary data modules to the 2.0 terminology (prefixes.py → particles.py), with a 2.x deprecation bridge #293's freeze exempts REGEXES (test_every_vocabulary_constant_is_frozen names it). customize.rst spends a paragraph explaining that editing it does nothing.

Proposed

Remove nameparser/config/regexes.py in 3.0:

• Rework test_regex_sync.py to pin each hand copy to the pipeline's copy directly (_render._ROMAN to _vocab._ROMAN, and so on). Entries already mapped to None there are unaffected.
• Drop regexes.REGEXES from test_contracts.py's exempt mappings, and the module from docs/modules.rst's automodule list.
• Add a row to migrate.rst's removals saying a caller who reused a pattern should copy it.
• Update customize.rst's freeze paragraph, which then names CAPITALIZATION_EXCEPTIONS alone.
• Reword the "Ported from v1 (nameparser/config/regexes.py ...)" provenance comments in _vocab.py, _render.py and _tokenize.py so they don't point at a file that no longer exists.

Open

• This reverses part of a recorded decision. decisions/3-0-reevaluations's 2026-10-28 Remove the 1.x vocabulary names (PREFIXES, FIRST_NAME_TITLES, …) and their bridge #357 entry lists regexes among the nameparser.config modules that stay in 3.0. Removing it needs a superseding entry there, as would anything Should from nameparser.config.titles import TITLES still work in 3.0, now that Lexicon.readings_of() answers what a word is? #1007 decides.
• 2.8 gave no warning. 3.0's other removals were announced by 2.8's deprecation warnings; this one was not. The options:
  • Remove it in 3.0 with the migrate.rst row. Callers who import a pattern for their own code break, and each fix is a paste.
  • Warn on import through a module __getattr__ in 3.0 and remove it in 4.0. This keeps the third copy and the backwards test anchor for a whole major version.