feat(adapter/junie): JetBrains Junie skill runtime

**Goal:** Bring JetBrains Junie to feature parity with Claude Code as a skill runtime. Grounds: [RFC-AI-0004 Principle 3 — Vendor neutrality](../tree/main/docs/rfcs/RFC-AI-0004.md).

**Why Junie:** JetBrains IDEs cover a large slice of the Java / Kotlin / PyCharm / IntelliJ user base — that includes most of the Apache Java-stack PMCs (Lucene/Solr, Cassandra, Camel, Beam, Flink, Kafka, Pulsar, NetBeans, …). Reaching parity means a maintainer can run framework skills from inside their JetBrains IDE without context-switching to a separate terminal CLI.

**What "parity" means:**

- Skills under `.claude/skills//SKILL.md` are invokable from Junie
- The `tools/*` bridges are reachable — Junie supports shell command execution
- Sandbox parity: Junie runs inside the JetBrains IDE process; document threat-model considerations and the mapping to the framework's sandbox baseline (different — IDE-hosted)
- HITL primitives: Junie has per-action confirmation natively — verify and document

**Suggested approach:**

- Map `SKILL.md` → Junie's `.junie/guidelines.md` convention (Junie reads these natively as project-level agent instructions)
- Document skill-invocation: how a maintainer triggers (e.g.) `security-issue-triage` from inside the JetBrains IDE
- Validate which framework skills are well-suited to IDE-hosted execution vs better as terminal CLI flows

**Reference:**

- RFC-AI-0004: [`docs/rfcs/RFC-AI-0004.md`](../tree/main/docs/rfcs/RFC-AI-0004.md)
- Junie: https://www.jetbrains.com/junie/
- `.junie/guidelines.md` convention: https://www.jetbrains.com/help/junie/customize-guidelines.html
- Existing skill shape: [`.claude/skills/`](../tree/main/.claude/skills/)