Remove the 1.x vocabulary names (PREFIXES, FIRST_NAME_TITLES, …) and their bridge

[derek73]

#293 renamed the nameparser/config vocabulary to the 2.0 terminology in 2.2 and kept every 1.x name resolving through a DeprecationWarning bridge. The warning names 3.0 as the removal, so this is that commitment coming due.

What to delete

Verified against master at 5331b5a.

Whole files — these exist only to serve the bridge:

• nameparser/config/_deprecated.py (133 lines) — the alias_getattr factory
• nameparser/config/prefixes.py (38 lines) — shim, no data
• nameparser/config/bound_first_names.py (23 lines) — shim, no data
• tests/v2/test_config_aliases.py (327 lines) — the bridge's entire test surface

In-module blocks, both at the end of the file after the import-time asserts:

• nameparser/config/titles.py — the TYPE_CHECKING/alias_getattr pair, and FIRST_NAME_TITLES from __all__
• nameparser/config/suffixes.py — the same pair, and SUFFIX_NOT_ACRONYMS from __all__

Both __all__ lists themselves stay: they are load-bearing for autodoc_member_order = 'bysource' and must remain in source order. Removing only the TYPE_CHECKING branch and leaving the retired name in __all__ is a ruff F822 error rather than a silent one, so the tree will tell you if you do half the job.

Prose:

• docs/migrate.rst — the bridge table and the section around it (~lines 236-260), plus the flip-warning reference at ~411
• AGENTS.md:224 — the retired-name half of the frozen-constants gotcha; the freeze half stays
• Leave docs/release_log.rst alone entirely. Its 2.2 entry is the historical record of what shipped.

Decide first: where the vocabulary lives in 3.0

The migration spec says 3.0 removes nameparser.config "in its entirety" while its parenthetical enumerates only the five v1 shim names (Constants, CONSTANTS, SetManager, TupleManager, RegexTupleManager). Those disagree, and #293 gave the disagreement a cost: Lexicon's public field docstrings now cross-reference nameparser.config.particles by path, and docs/modules.rst autodocs the config package. If 3.0 moves the vocabulary under the core (say nameparser/_vocabulary/), roughly nine :data: references and a docs section move with it — and the module name #293 established changes address one version later.

Settle that before the sweep rather than during it. The answer decides whether this issue is "delete a bridge" or "delete a bridge and relocate the vocabulary".

If it does move, note what #293 learned the hard way: git records no renames, it infers them at diff time by pairing a deleted path with an added one, so the move must be its own commit with nothing re-created at the old path, or git blame on the vocabulary restarts at the move. See PR #354's e4f8385/9956e7d split.

Acceptance

• grep -rn "PREFIXES\|NON_FIRST_NAME_PREFIXES\|BOUND_FIRST_NAMES\|FIRST_NAME_TITLES\|SUFFIX_NOT_ACRONYMS" nameparser/ docs/ AGENTS.md returns hits only in docs/release_log.rst
• from nameparser.config.prefixes import PREFIXES raises ModuleNotFoundError
• from nameparser.config.titles import FIRST_NAME_TITLES raises ImportError
• no DeprecationWarning remains anywhere in nameparser/config/
• suite, mypy, ruff and the sphinx build green with the four files gone