chore(release): v0.1.18 — rebalance hydration philosophy

[HenryLach]

Changes

Rebalances the hydration methodology across the skill, worker agent, and reviewer agent to favor outcome-level checkboxes over exhaustive implementation scripts.

Problem

Wave 1 tasks (TP-005, TP-006, TP-010) averaged ~65 min each. Analysis showed the biggest time sink was excessive plan hydration: workers expanding 2-bullet steps into 15+ micro-checkboxes, then reviewers demanding even more function-level detail before approving. The review system was catching real bugs, but the plan-hydration cycle was adding ~15-20 min per step.

Solution

• Skill v1.1.0 (skills/create-taskplane-task/SKILL.md): Rewrote hydration section with 'Adaptive Planning, Not Exhaustive Scripting' philosophy, anti-pattern callouts, good/bad granularity examples
• Worker agent (templates/agents/task-worker.md): 'One checkbox per meaningful outcome' replaces 'one per unit of work'; added calibration heuristic
• Reviewer agent (templates/agents/task-reviewer.md): Added Plan Granularity Guidance section — ask for missing outcomes/risks/test intent, not function-level implementation checklists
• Prompt/STATUS templates: Updated hydration banners
• Pending tasks (TP-007, TP-009, TP-012): Updated banners; TP-007 Step 0 collapsed from 8 micro-checkboxes to 3 outcome-level items

Validation

• 290/290 tests pass on feat/polyrepo-support (the shipped extension code is unchanged)
• Pre-existing test failures on main (4 files) are unrelated to this change
• CLI smoke: taskplane help + taskplane doctor clean
• npm pack --dry-run confirms correct package contents