Avoid pod restarts on global config changes where possible (e.g. queues updates)

Currently the mechanism added in HPCC-24967 means that pods update whenever their configuration changes.
That includes parts of their configuration which are global in nature, e.g. the list of target queues.
It would be good if these can be excluded from the auto-restart mechanism (the config SHA annotation) and instead spotted by existing running pods, and reloaded internally.