AMQP: Connection failure at startup should trigger async connection attempt - not fatal error
Currently the backend does not start if the connection to the AMQP broker fails (in case an AMQP host (=broker) is configured). Even though the AMQP functionality is a core element of the VILLASweb backend, a fatal error is probably too harsh as a consequence here.
In case of an AMQP connection failure on startup, I propose to run an async thread that attempts to connect to the broker in intervals of X seconds until it is connected successfully. That way, at least the non-AMQP endpoints of the API would be available, even if there is an AMQP issue in the system.
Of course, we would need to make sure that all endpoints and functions that use AMQP, test if a connection exists prior to sending anything and return an error code if the connection is not available.