Non-admin users (and who don't have the cluster-monitoring-view role) can use netobserv thanks to Loki multi-tenancy, however that doesn't exist with prometheus. When both prom & loki are enabled, this is ok because they can choose to force using the Loki datasource (prom would return a 403)

However, this isn't very convenient: not only because the "auto" mode uses prometheus when possible, hence showing an error; but also because, when they force it to Loki, this setting isn't memorized, e.g. when leaving the Network Traffic page then going back to it, it shows again the prom error.

1. Install netobserv w/ loki operator & prometheus enabled
2. Setup multi-tenancy as described here: https://github.com/netobserv/documents/blob/main/loki_operator.md#testing-multi-tenancy 
3. Navigate to Topology (using a query compatible with prometheus) => error
4. Switch to using Loki => it works
5. Navigate to another page of the OCP console, then come back to netobserv => The Datasource setting is back to auto, hence showing an error

Errors displayed by default, and Datasource setting not preserved

Ideally, it should use Loki by default if we can pre-detect that prom queries would fail.

At least, the datasource setting should be memorized.