Add local-up.sh

[llhuii]

Developers can run hack/local-up.sh to setup up a local environment
including:

1. a local k8s cluster with a master node.
2. a kubeedge node.
3. our gm/lc.

Based on the kubeedge-local-up script which builds a local k8s cluster
and kubeedge, our local-up script installs our package locally for
simply developing and preparing for e3e tests.

It does:

1. build the gm/lc/worker images.
2. download kubeedge source code and run its localup script.
3. prepare our k8s env.
4. config gm config and start gm.
5. start lc.
6. add cleanup.

For cleanup, it needs to do our cleanups before kubeedge cleanup
otherwise lc cleanup (via kubectl delete) is stuck and lc container is
kept running.

[llhuii]

wait this kubeedge/kubeedge#2495 to merged.

[llhuii]

Now since we builds images locally, we only can start gm at edgenode(at which edgecore running locally).

TODO: deploy gm to test-control-plane master node which are created by kind .

Found a way to deploy gm image to test-control-plane : kubernetes-sigs/cri-tools#546.

docker save localhost/edgeai-neptune/neptune/gm:localup | docker exec -i test-control-plane ctr --namespace k8s.io image import - 

[llhuii]

with setsid command, we can handle our cleanups before cleanup kubeedge.

[llhuii]

docker save localhost/edgeai-neptune/neptune/gm:localup | docker exec -i test-control-plane ctr --namespace k8s.io image import - 

just use kind load docker-image command.

[llhuii]

1. updated with some updates merging Optimize installation docs #48.
2. add neptune-lc manually clean in case edgecore terminated unexpected.