Currently, cli uses a custom tcp exporter similar to the one implemented for packets and rely on oc port-forward command to reach the eBPF pods.

We should try to reuse gRPC exporter instead to simplify the agent code and see if we can also optimise the packet approach.

Since FLP Direct mode is used for flows, we should also consider adding a stage for that.

This task may need to deploy the cli in a pod on the cluster and run oc exec instead.

The output files can be copied before leaving the command.

See:

https://github.com/netobserv/network-observability-cli/pull/1/files#diff-d60408a7478e2216d3b844cf49af5b3372826538b10ddf883c3ad9ae3d79619cR46 

https://github.com/netobserv/network-observability-cli/pull/1#discussion_r1397326137 

https://github.com/netobserv/netobserv-ebpf-agent/pull/224#issuecomment-1816169973