Refactor Apiserver e2e run in cluster

[machichima]

• Create a curl container in the same pod as apiserver for sending curl request
• Add make command to start up pod with both apiserver and curl for e2e test
• Create exec.go for sending request to apiserver through remote command (similar to kubectl exec behaviour).
  • The exec.go contains all functions in pkg/http/client.go, but send through remote command

Why are these changes needed?

Currently, the KubeRay API server e2e tests rely on accessing the API server via a NodePort exposed through kind's extraPortMappings, defaulting to http://localhost:31888. This approach depends on the local environments. Everything should run inside the Kubernetes cluster.

Related issue number

Closes #3380

Checks

• [ ] I've made sure the tests are passing.
• Testing Strategy
  • [ ] Unit tests
  • [ ] Manual tests
  • [ ] This PR is not tested :(

[machichima]

@dentiny @rueian PTAL. Thanks!

[machichima]

Hi @rueian @dentiny , The autoscaler e2e test is failing, which I think is flaky. Could you please re-trigger it for me?

Thanks!

[rueian]

Hi @rueian @dentiny , The autoscaler e2e test is failing, which I think is flaky. Could you please re-trigger it for me?

Thanks!

Hi @machichima, I don't have the permission to retrigger it. Perhaps you could merge the master branch to retrigger it.

[machichima]

@dentiny , PTAL Thanks!

[machichima]

Hi @kevin85421, Could you please have a look at this?

[kevin85421]

@machichima can you resolve the conflict? @rueian can you shepherd this PR and merge if you think it is good to go? Thanks!