fix: remove hardcoded infra details, tighten e2e CI, drop validation report docs

[Copilot]

Public repo contained environment-specific secrets/identifiers in committed files, and the e2e CI job was failing hard on the openclaw-not-on-npm issue instead of continuing gracefully.

Security: scrub personal/infra details

• docs/DEPLOYMENT-SOP.md: Replace all deployment-specific references with generic placeholders — ssh zeph@revere → ssh user@your-host, /share/Container/ → /path/to/containers/, named instances (Zephyr, Seraphel) → my-bot/my-advisor, removed Apple Notes/iCloud/Tailscale/QNAP, "BCM server" → "internal operator/monitoring Discord server", dropped "Todd/CHC (Certainty Home Consulting)" attribution. Fixed volume mount paths to match actual Dockerfile (/app/workspace → /home/openclaw/.openclaw).
• scripts/fleet.sh snapshot command: Replace hardcoded DOCKER_HOST=tcp://192.168.1.2:2376 and DOCKER_CERT_PATH=$HOME/.docker/nas defaults with an explicit guard — operator must export all three Docker TLS vars before calling snapshot:

  if [[ -z "${DOCKER_HOST:-}" || -z "${DOCKER_TLS_VERIFY:-}" || -z "${DOCKER_CERT_PATH:-}" ]]; then
    echo "Error: DOCKER_HOST, DOCKER_TLS_VERIFY, and DOCKER_CERT_PATH must be set..." >&2
    exit 1
  fi

CI: graceful e2e failure

• .github/workflows/test.yml: Add continue-on-error: true to the e2e-test job, consistent with how ci.yml handles the Docker build step. The e2e script already exits 0 on known npm-unavailability; this covers runtime health-check failures when no API keys are present.

Cleanup

• Remove DOCKERFILE_VALIDATION.md and ISSUE_3_VALIDATION.md — noise, not docs.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.