Bridge create-learning-path to LPs suggested from learning journey creation

## Summary

Update `/create-learning-path` so authors can either start a path from scratch **or** continue from a learning path that was **suggested during learning journey (LJ) creation**. Today the command always assumes a greenfield feature/audience prompt and does not reuse analysis or cross-link back to the source journey.

## Problem

When LJ authoring suggests a follow-on learning path, that suggestion (and the analysis behind it) is easy to lose. `/create-learning-path` has no first-class entry point for “this path was suggested from an LJ,” so authors re-explain context, re-do research, and often ship a path that never links back to the journey that recommended it.

## Proposed behavior

### 1. Entry prompt on invoke

When `/create-learning-path` starts, ask the author which mode they want:

1. **New topic** — what should this learning path teach? (current behavior: feature/product goal + audience)
2. **Suggested from a learning journey** — continue a path that LJ creation already recommended

### 2. Suggested-from-LJ flow

If the author chooses option 2:

1. Prompt for the **learning journey PR URL** (or number) in `grafana/interactive-tutorials`.
2. Locate the **architectural / analysis artifact** produced by the LJ creation skill for that work (exact path TBD — to be filled in once identified; leave a clear “look here first” note in the skill once known).
3. **Import** the relevant context and analysis from that artifact (topic framing, audience, why a path was suggested, milestone ideas, docs already reviewed, constraints, etc.) instead of starting from a blank feature prompt.
4. Proceed through the normal create-learning-path phases (propose milestones → scaffold → manifests → selectors → test → wrap-up), using the imported context as the default plan the author can refine.

### 3. Cross-link the journey from the new path

The generated learning path should reference the source journey as a related resource, for example via:

- `related_journeys` / “Related paths” / “See also” / additional-resources style content in `content.json` and/or `website.yaml`
- Manifest linkage where appropriate (`suggests` / `recommends` / side or related journeys), consistent with existing LP packages

Goal: someone finishing (or discovering) the new path can find the original journey, and vice versa where that already exists on the LJ side.

## Acceptance criteria

- [ ] `/create-learning-path` opens with a mode choice: **new topic** vs **suggested from LJ**
- [ ] Suggested mode requires an LJ PR link and fails clearly if the PR/artifact cannot be resolved
- [ ] Skill imports analysis from the LJ architectural artifact (path documented in the skill once confirmed) and does not ask the author to restate that analysis from scratch
- [ ] Generated path includes a see-also / related / additional-resources link back to the source journey
- [ ] Greenfield (new topic) flow remains available and unchanged in intent
- [ ] `.cursor/commands/create-learning-path/README.md` (and related workflow docs if needed) describe the bridge

## Open questions / follow-ups

- Exact path and schema of the architectural analysis file from LJ creation — **author will identify and add to this issue**
- Whether the LJ PR must already be merged, or draft/open PRs are enough to read artifacts from the branch/workspace
- Bidirectional linking: should this skill also propose a follow-up edit on the LJ to `suggests` the new path ID once it exists?

## References

- `.cursor/commands/create-learning-path/README.md`
- `.cursor/learning-path-workflows/workflows.md`
- Existing related-content patterns: `related_journeys`, `side_journeys`, path/step `suggests` / `recommends` in `docs/manifest-reference.md` and `docs/website-yaml-reference.md`