Indexing is slow in OpenShift

[yrodiere]

I had to create #372 to try to workaround the issue, which might have been hidden for weeks by the problem I solved in #371 ...

Essentially indexing is very slow, taking much more than the 240s it used to take.

This might be caused by a change of hardware in the OpenShift cluster, in which case we're f*ed so I'd rather explore other possible causes.

This might be caused by the switch to Elasticsearch, so we would need to check the configuration -- did we forget anything when we switched over from OpenSearch?