Using the OOTB flowcollector CRD, the ebpf flow collectors repeatedly CrashLoop for reason OOMKill under a very modest network load.

1. AWS cluster withh 9 m5.2xlarge workers
2. Install NO with its default 100Mi memory limit in the flowcollector CRD
3. run the hey-ho app (https://github.com/jotak/hey-ho) with 10 projects, 10 deployments and 1 replica
4. oc get pods and watch the netobserv-ebpf-agent pods CrashLoop

The network traffic per node is 20K-300K flows/minute and roughly 200Mb/s spread for 1-2 pods per node.

We should remove the memory limit for the collector unless we know a correct limit for our target flow and traffic rates.   The OOTB default should not crash so easily.