Kudos @meks007 for providing a solution for #158 - during the start of the Integration it will be checked, if the configured evcc server is reachable and raise ConfigEntryNotReady exception, if this is not the case.
Your HA installation should (& will) handle this early ConfigEntryNotReady signal appropriate by retry to initialize the Integration again (after a short while) - this loop will be repeated till the integration will finally can reach the evcc server.
Local tests have shown that this HA-retry-to-init-integration-loop can run for over 1h... where the pause to the next retry will increase over the attempts.
As so often - just let HA handle the situation and "just" provide the correct signals. I am not sure where I have left this path with this evcc-integration. Nevertheless once more Thanks @meks007.
Ceterum censeo Carthaginem esse delendam - I am still have the opinion, that (automated) frequent re-starts of a HA-Server can be and appropriate solution to any problem.