Objective-first task-step semantics across superRA workflow

[FuZhiyu]

Summary

• make task headings plus Script / Input / Output the scope contract, with steps treated as planner-authored starter guidance across contributor docs, handoff-doc canon, agent contracts, workflow skills, and the data-analysis discipline
• require within-task route-rewrite rationale to live in the implementer status message layer rather than PLAN.md history
• add focused Claude Code coverage for the new objective-first semantics, make the shared test harness portable on macOS, and retire the stale subagent-driven-development script from the default fast suite
• archive the finalized handoff docs under docs/plans/2026-04-22-objective-first-task-step-semantics-{plan,results}.md

Verification

• cd tests/claude-code && ./run-skill-tests.sh
• python3 skills/codex-superra-setup/scripts/sync_codex_agents.py --scope project --check

Notes

• docs/plans/2026-04-22-objective-first-task-step-semantics-results.md contains the permanent record and archived verification logs.
• This branch was integrated against current origin/main during Phase B before PR creation.