[richie] re-generate ES index after rolling update

[jmaupetit]

Purpose

When for some reason, Richie's app ES pod is re-deployed (rolling update), the search index required to query ES (from Richie's catalog view) no-longer exists, raising an HTTP 500 error code.

Proposal

One might consider adding a new volume to persist ES data, but we are afraid of a strong performance impact. This should be tested/considered.

An alternative would be to use a post deployment hook to generate this index from Richie's management command. In this case, we should consider that Richie + ES are containers from the same pod. As it would impact application scalability, I'm also not confident with this solution.

We are wide open to suggestions.