When an agent establishes a connection with FLP service, it keeps that connection the whole time, without rebalancing even when FLP is scaled up. As a result, scaling up FLP results in spawning new pods that aren't used at all.

Investigate how we can fix that (Service sticky session config? iirc we used to have that in the early days with IPFIX; Alternatively, the operator could watch for FLP being scaled-up, and force a rebalance by restarting agent pods, but that's not very nice)

See also: https://github.com/netobserv/network-observability-operator/pull/1953#issuecomment-3529216041