[product-manager-impersonator] /ai help fast-path for mention workflows

## Feature Idea

**Summary:** Add a built-in `/ai help` fast-path in mention workflows so bare `/ai` (or `/ai help`) returns a concise capability menu with example prompts instead of immediately entering a full free-form agent run.

## Why a Customer Would Want This
New adopters see "Type `/ai`" in docs, but often do not know what to ask first. A first-run help reply lowers trial-and-error, reduces noisy follow-up comments, and makes the workflow feel more predictable for maintainers onboarding teams.

## Rough Implementation Sketch
- In `.github/workflows/gh-aw-mention-in-issue.md` and `.github/workflows/gh-aw-mention-in-pr.md`, add an instruction branch: when `steps.sanitized.outputs.text` is empty, `/ai`, or `/ai help`, return a short command-style help response.
- Keep using the existing safe output comment paths (`safe-output-add-comment-issue` and `safe-output-add-comment-pr`) so no new write surface is introduced.
- Add 4-6 task-oriented starter prompts (question answering, PR review, bug fix, follow-up context) tuned separately for issue vs PR contexts.
- Document the new behavior in `gh-agent-workflows/mention-in-issue/README.md` and `gh-agent-workflows/mention-in-pr/README.md`.

## Why It Won't Be That Hard
The trigger and reply plumbing already exist, and request text is already sanitized and injected into the workflow prompt context. This is primarily a small prompt-routing and docs update across two workflow sources plus their READMEs.

## Evidence
- `docs/index.md:34` advertises "Type `/ai` in any issue or PR" but does not provide in-band first-use guidance.
- `gh-agent-workflows/mention-in-issue/example.yml:18-19` and `gh-agent-workflows/mention-in-pr/example.yml:20-22` route any `/ai`-prefixed comment into the workflows.
- `.github/workflows/gh-aw-mention-in-issue.md:108` and `.github/workflows/gh-aw-mention-in-pr.md:144` already pass `steps.sanitized.outputs.text` into the assistant request context.
- `.github/workflows/gh-aw-mention-in-issue.md:14` and `.github/workflows/gh-aw-mention-in-pr.md:19` already include safe-output add-comment fragments that can carry a help reply.
- Recent active work in this area (`#2015`, `#2032`) shows mention and safe-output UX/reliability are current maintenance priorities.

## Suggested Actions
- [ ] Add `/ai help` detection and response behavior in mention workflow source files (`.md`) and recompile.
- [ ] Add concise help examples to both mention workflow READMEs.
- [ ] Add a brief docs note in `docs/index.md` clarifying that `/ai help` returns starter prompts.

---
[What is this?](https://ela.st/github-ai-tools) | [From workflow: Trigger Product Manager Impersonator](https://github.com/elastic/ai-github-actions/actions/runs/34587431017)

Give us feedback! React with 🚀 if perfect, 👍 if helpful, 👎 if not.

Start with .github/workflows/gh-aw-mention-in-issue.md and gh-aw-mention-in-pr.md, tracing steps.sanitized.outputs.text and the existing safe-output add-comment fragments. Add the specified empty, /ai, and /ai help behavior, then update both mention workflow READMEs and docs/index.md. Recompile the workflows and verify that issue and PR contexts each show their intended starter prompts.

Written by the indexing model from the issue text.