prometheus storage.tsdb.retention.size should be configured to match the PVC size

Summary

On my dev cluster, Prometheus pod recently failed in crashloop, hitting that error:

time=2026-02-11T09:45:30.144Z level=ERROR source=main.go:1431 msg="Fatal error" err="opening storage failed: open /prometheus/wal/00000408: no space left on device"

In order to prevent such issues, we could set configure prometheus.spec.retentionSize following sizing recommandations

It's indeed safer that time-based limitation, since the amount of data stored depend not only in the retention delay, but also on the amount of metrics that are collected.