fix: restore helm delete and increase resilience to network disconnects

[starpit]

we have a few places that are doing kubectl logs, without protection against network disconnects