Quickstart: reproduce & verify advancing-next board unification

Reproduce the bug (pre-fix, current main)

The #4980 standalone reproduction (also captured in this mission's tooling-friction/approach tracers) drives a 1-WP single_branch mission to a review rejection and asserts the loop spins:

# From the source checkout:
SPK_QA_SOURCE="$PWD" bash <the #4980 repro script> /tmp/repro_out
# Expect (pre-fix):
#   bug advancing next [a1..a3] rc=0 kind=step state=review action=review wp=None
#   bug query next rc=0 kind=query state=implement action=None wp=WP01
#   CONFIRMED: after a review-step rejection the loop spins on a WP-less placeholder review step (exit 0); controls behave

For #4975, create a mission with --topology coord and advance the loop to implement:

# Expect (pre-fix):
#   coord advancing next 5 rc=1 blocked implement action=None wp=None reason=No action mapped for WP step 'implement'
#   coord query next: query implement wp=WP01

Verify the fix (post-fix)

# Blast-radius test tier (NFR-004, ~<90s):
.venv/bin/python -m pytest tests/runtime/test_bridge_parity.py tests/next/test_finalized_task_routing.py tests/next/test_decision_unit.py -q
# Plus the advance-guard tests:
.venv/bin/python -m pytest tests/runtime/next/ -q

Post-fix expectations:

  • bug arm of the #4980 repro re-dispatches implement WP01 (no CONFIRMED spin); approve/early controls unchanged.
  • coord advancing next --result success dispatches implement WP01 (exit 0), matching query mode.
  • Every parity-matrix cell (incl. combined coord×review) green; blocked floor returns kind=blocked + recovery.

Dogfooding caveat

This mission is itself coord topology. Do NOT drive its own implement/review via the buggy autonomous spec-kitty next --result success loop (it would hit #4975/#4980). Use spec-kitty implement WP01 directly and orchestrate review manually.