Provincial run autonomy: rename scripts to noun_verb + single 'approve once, builds end-to-end' wrapper

[NewGraphEnvironment]

Problem

After today's session (PR #171, v0.36.1) the operational scripts work but are scattered, inconsistently named, and require operator handholding mid-run. We want a single command — approved once — that runs end-to-end and lands clean output. M4+M1 only as the validated baseline; cyphers as opt-in after the baseline lands repeatably.

Current naming violations

The link script family is noun_verb (per NewGraphEnvironment/soul#46, matches rtj's cypher_up.sh / cypher_down.sh / cypher_run.sh). Several existing scripts violate this:

Current	Convention	Proposed rename
trifecta_provincial.sh	noun_verb	wsgs_dispatch.sh
run_provincial_parity.R	noun_verb	wsgs_run.R
compare_bcfishpass_wsg.R	noun_verb	wsg_compare.R
consolidate_schema.R	noun_verb	schema_consolidate.R
archive_provincial_runs.sh	noun_verb	runs_archive.sh
balance_provincial_buckets.R	noun_verb	buckets_balance.R
province_run.sh	(already correct)	unchanged
province_clean.sh	(already correct)	unchanged
province_progress.sh	(already correct)	unchanged

Plus the names lie about scope — these scripts run any list of WSGs, not just "provincial".

Goals

1. Single-command autonomous run. Operator approves bash data-raw/<umbrella>.sh ... once; everything inside (state-clean → snapshot → dispatch → pull → consolidate → burn cyphers if any) runs without further prompts.
2. Any WSG list. --wsgs=A,B,C accepted at the umbrella level, auto-split via LPT across configured hosts.
3. Any config bundle. --config=default or --config=bcfishpass (default), --schema=<name> for output schema.
4. Any host subset. --no-cyphers (M4+M1 only) for the validated baseline; --cy-workspaces=job1,job2,job3 for full distributed.
5. Rename for honesty. No more "provincial" / "trifecta" / "bcfishpass" in script names that work for any list/host count/reference.

Acceptance

• Phase 1 baseline: bash data-raw/province_run.sh --wsgs=<16-WSG-test-set> --config=default --schema=fresh_default --no-cyphers --with-mapping-code runs ~30 min wall on M4+M1, lands 16 WSGs in fresh_default.streams on M4, no operator prompts, exit code 0.
• Phase 2 cypher integration: same command with --cy-workspaces=job1 (single cypher) runs end-to-end + burns the cypher on completion.
• Phase 3 full distributed: same command with --cy-workspaces=job1,job2,job3 runs full 217-WSG provincial.
• All scripts renamed per noun_verb convention; git log --follow works (renames are pure renames, not delete+add).
• Internal references updated (source() calls, doc references, README, runbook, PWF, post_compact handoff).
• Tests pass (devtools::test()).
• bash -n syntax-clean on all 4 shell scripts.

Out of scope (separate issues)

• Decouple compare from pipeline (Decouple bcfp compare from link pipeline run — separate compare-only step #168)
• Autossh for tunnel stability (bcfp tunnel drops during long provincial runs cause silent per-WSG errors #167)
• S3-based consolidate (S3-based consolidate: route pg_dumps through s3://newgraph/ instead of serial scp #170)
• Simplify lnk_persist_init (Simplify lnk_persist_init after clean cypher snapshot lands #169 after rtj#145)
• M4 PG over-tuning (fresh#199)

Reference

• Today's session learnings: planning/active/findings.md (12 gotchas) + progress.md
• Convention reference: NewGraphEnvironment/soul#46
• The 16-WSG test set: CARP, CRKD, FINA, FINL, FIRE, FOXR, INGR, LOMI, MESI, NATR, OSPK, PARA, PARS, PCEA, TOOD, UOMI