strongSwan start_action=start only fires at config load; after a rekey
failure the SA dies and is not re-initiated automatically. This caused the
divserver (telemt MTProxy) tunnel to stay down for days after a failed
rekey on 2026-07-14.

Add a systemd timer (every 2 min) that re-initiates down SAs on parentglobal:
  - divserver/div   -> divserver (telemt MTProxy Greek egress)
  - site-to-site/s2s -> ikev2.gr.egw/.139 (gvpn Greek egress)

Mirrors the fr-tunnel docker autoheal, but on systemd. README documents
deployment via reverse-SSH (anyssh.ru:10337) and verification.
Co-Authored-By: Claude <noreply@anthropic.com>