fix(docker): clear stale dind state on startup

[Canvinus]

Summary

• clear stale inner-Docker pid/socket files before starting dockerd
• replace docker stats startup probe with docker info
• exec Nest after dockerd is ready

Incident context

Prod returned 502 after the host rebooted on 2026-07-08 19:47 UTC. nginx and the host were healthy, but main_sscan-nest never launched Nest because dockerd failed with:

failed to start daemon, ensure docker is not running or delete /var/run/docker.pid: process with PID 8 is still running

Manual prod recovery removed stale /var/run/docker.* state and restarted only main_sscan-nest; both api.sourcescan.dev and api-v2.sourcescan.dev returned 200 after that.

Verification

• sh -n startup command syntax check passed
• git diff --check passed
• prod recovery verified the same stale-state cleanup path live: inner Docker 29.5.3 responded, local Nest/nginx returned 200, public APIs returned 200

Note: local privileged docker:dind runtime test was not run because privileged dind on the workstation was blocked by safety policy.