Make /healthcheck endpoint capture health of OPA connection

[v-rocheleau]

Background

Authz needs a healthy connection to OPA to function correctly, issue #64 breaks this and is still being investigated.

The temporary solution to get the deployment healthy again is to restart the authz deployment in k8s.

Issue

The authz deployment has probes on authz's /healthcheck endpoint, if that endpoint returns non 200 status k8s is able to restart the deployment automatically.

Problem: the /healthcheck endpoint keeps returning 200 statuses even if the connection to OPA is unhealthy, preventing automatic re-deployment and forcing operators to restart the deployment manually, after outages are noticed by submission users.

Note

A definitive solution to #64 should still be implemented, this issue is strictly to remove the need for manual interventions when the issue occurs.

[daisieh]

I've deployed a temporary check to at least log the issue if it happens, but I'm loath to actually have it return "unhealthy" to trigger restart until we have persistent logging on the environments.