add container resource profiles

[willcl-ark]

Closes: #26

Adds a few demo profiles such as laptop, desktop, raspberry_pi, server to the graph config.

While we are using docker-compose, and not swarm, we cannot easily limit physical CPUS (withough downgrading to docker-compose v2), but we can get "relative resourse usage", and memory limitations.

This line:

"cpu_shares": int(self.resource_profile.get("cpus", "0.5")) * 1024,

is the temporary "hack" which converts our cpu number from a profile to a rough cpu_shares value (which docker-compose accepts).

Link: https://docs.docker.com/config/containers/resource_constraints/

[josibake]

Concept ACK

Need to look at this with the k8s backend, but I think it should work by having the resource profiles and then each backend translates that into something the backend can understand

[willcl-ark]

If we want ot continue with tc_netem, then I think it just needs a big rebase.

If we want to pass it as actual pod resource profile to the cluster (can you even do that), then it might need a total re-write...

[josibake]

You can set pod resource limits (CPU, memory) directly when defining the pod in k8s. But we could also use tc_netem throughout, because if I remember correctly that's happening inside a docker container? So it should work on both and iirc, it gives more options than just CPU/RAM.

Maybe we just stick with tc_netem through for now? Longterm, I can see value in having the profile be a "warnet" concept, and then each backend translates the profile into what the backend needs