[bug] High operator memory in large clusters

[jdheyburn]

I deployed 0.1.0 to a non-prod cluster that has about 8000 resources in it (pods, configmaps, etc.) and the operator memory usage was about 900-1024MiB despite there being no managed objects being deployed.

We should keep the operator running lean, no matter what the size of the cluster it is deployed in.

[bjosv]

Maybe we can disable caching for sensitive data like secrets, like cloudnative-pg does,
and maybe only cache ByObject that are managed by the operator (label app.kubernetes.io/managed-by: valkey-operator)

[jdheyburn]

@bjosv I'm for both of those ideas, I raised #178 as an option too. I saw CNPG doesn't do this though, so we can deprioritse it and try the other 2 solutions you mentioned.

[deepakpunjabi]

Looks similar problem as #117.