feat: add /escalation skill — incident triage and war room coordination

[HMAKT99]

3am. Production is down. Who do you page?

The difference between a 30-minute outage and a 4-hour outage is almost never technical. It's knowing the severity, who to escalate to, and whether to fix forward or rollback. Most teams figure this out during the incident. /escalation figures it out before.

What /escalation does

You:   /escalation --incident "Dashboard returning 500 errors"

Claude: SEVERITY: SEV-2 (HIGH)
        Blast radius: Dashboard users (~40% of traffic)
        Active: Yes — errors ongoing

        ESCALATION PATH:
        TIER 1 → alice (most active on app/controllers/)
        TIER 2 → bob (backend lead, last 30 days)

        ROOT CAUSE ANALYSIS:
        Most likely: commit abc123 (2 hours ago)
        "Add caching to dashboard queries"
        Evidence: 3 files changed in app/services/dashboard/

        Fix forward or rollback?
        RECOMMENDATION: A) Rollback — clear causal commit, users impacted now

SEV-1 through SEV-4 classification. 4-tier escalation with auto-detection of owners from git history. War room playbook (IC, Tech Lead, Comms, Scribe roles). Post-incident review with 5 Whys.

Only .tmpl committed — bun run gen:skill-docs generates the rest.

Test plan

• .tmpl follows template pipeline
• Registered in gen-skill-docs.ts, skill-check.ts, test files