Handle Failures and Parallelization of calls to the Clusters

[luizbafilho]

Currently, we loop through all clients to all clusters when listing resources https://github.com/weaveworks/weave-gitops/blob/main/core/server/kustomization.go#L25

Given this is done serially it's not very efficient and should be done in parallel to optimize execution.

While doing the parallelization work we should also make sure a failure while calling a single cluster does not exit out the call and keep going until all clusters have been called.