Improve next-steps prompts with numbered options and context-clearing guidance

[williamthorsen]

Problem

The next-steps prompts produced by plan- and review-producing skills use ▶/· markers but no numbers. Selecting an option requires copy-pasting the full line or typing the skill name. There's also no guidance on when to clear context before proceeding.

Context

• The ▶/· format was introduced in #320 Add next-steps resumption prompt to plan-producing skills #329 via _data/next-steps-after-plan.md
• A second shared data file _data/next-steps-after-review.md was added in Modify review-change skill to check code against ticket, if one is given #328
• Both files define the format and recommendation rules that consuming skills reference
• Recommendation rules in next-steps-after-plan.md were tuned in next-steps-after-plan rules over-recommend refinement for interactive plans #351

Solution

Numbered options with emojis

Replace ▶/· markers with sequential numbers and consistent emojis. Bold only the recommended option's label.

Emoji assignments:

• 🧠 Refine plan / Design and plan
• 🎶 Orchestrate
• 🚀 Implement directly
• 📝 Update ticket

Example (next-steps-after-plan.md):

Next steps:
  1. 🧠 Refine plan: Clear context and use the `refine-plan` skill with plan: {plan_path}, ticket: {ticket_source}
  2. 🎶 **Orchestrate** (🟢 recommended): Clear context and use the `orchestrate-dev` skill with plan: {plan_path}, ticket: {ticket_source}
  3. 🚀 Implement directly

Context-clearing guidance

Prepend "Clear context and use..." vs "Use..." based on the option:

• Orchestrate — always "Clear context". Subagents start fresh; prior conversation wastes tokens.
• Refine plan / Design and plan — "Clear context". The plan artifact is self-contained.
• Implement directly — no "Clear context". Conversation history is valuable for manual implementation.
• Update ticket — no "Clear context". Ticket update benefits from review discussion.

Paths

Artifact paths in the output should use ~/-relative format when possible and absolute paths otherwise.

Acceptance criteria

• Both _data/next-steps-after-plan.md and _data/next-steps-after-review.md use numbered options
• Paths to artifacts use ~/-relative format where possible and absolute paths otherwise
• Each option includes context-clearing guidance where appropriate
• A consistent emoji is assigned to each option across both files
• Only the recommended option's label is bolded
• Recommendation rules in both files are unchanged (no functional changes to when each option is recommended)
• All consuming skills inherit the new format via the shared data files — no per-skill changes needed