tsh kube login by name prefix, labels, and/or query

[GavinFrazar]

This PR allows a user to select the current context by prefix name during tsh kube login, and to filter the kube clusters for login by --labels and/or predicate --query - these flags are not allowed with --all.

changelog: tsh kube login now supports --query and --labels optional arguments.

Part of the implementation for RFD 129.

As part of the changes, I also reduced the number of API calls to list kube clusters from 3 to one - in the case of a relogin required, or an access request is auto-generated, we of course still incur unavoidable extra API calls.

Example

$ tsh kube ls
Kube Cluster Name Labels                                                         Selected
----------------- -------------------------------------------------------------- --------
test              account-id=123456789012,env=fake,origin=gavin,region=us-west-1
test-cluster      account-id=123456789012,origin=gavin,region=us-west-1

$ tsh kube login --labels env=fake
Logged into Kubernetes clusters:
test

Select a context and try 'kubectl version' to test the connection.

$ tsh kube login --query 'labels["region"] == "us-west-1" && labels["env"] != "fake"'
Logged into Kubernetes clusters:
test-cluster-eks-us-west-1-123456789012

Select a context and try 'kubectl version' to test the connection.

$ tsh kube ls
Kube Cluster Name Labels                                                         Selected
----------------- -------------------------------------------------------------- --------
test              account-id=123456789012,env=fake,origin=gavin,region=us-west-1
test-cluster      account-id=123456789012,origin=gavin,region=us-west-1

$ tsh kube login test-clus
Logged into Kubernetes cluster "test-cluster-eks-us-west-1-123456789012". Try 'kubectl version' to test the connection.

$ tsh kube login test
Logged into Kubernetes cluster "test". Try 'kubectl version' to test the connection.

$ tsh kube ls
Kube Cluster Name Labels                                                         Selected
----------------- -------------------------------------------------------------- --------
test              account-id=123456789012,env=fake,origin=gavin,region=us-west-1 *
test-cluster      account-id=123456789012,origin=gavin,region=us-west-1

$ tsh kube login test-clus
Logged into Kubernetes cluster "test-cluster-eks-us-west-1-123456789012". Try 'kubectl version' to test the connection.

$ tsh kube ls
Kube Cluster Name Labels                                                         Selected
----------------- -------------------------------------------------------------- --------
test              account-id=123456789012,env=fake,origin=gavin,region=us-west-1
test-cluster      account-id=123456789012,origin=gavin,region=us-west-1          *

$ tsh kube login te
ERROR: kubernetes cluster "te" matches multiple kubernetes clusters:

Kube Cluster Name                       Labels                                                                                                                                                                                               Selected
--------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- --------
test                                    account-id=123456789012,env=fake,origin=gavin,region=us-west-1,teleport.dev/cloud=AWS,teleport.dev/origin=dynamic
test-cluster-eks-us-west-1-123456789012 account-id=123456789012,origin=gavin,region=us-west-1,teleport.dev/cloud=AWS,teleport.dev/origin=cloud,teleport.internal/discovered-name=test-cluster,teleport.internal/discovery-group-name=aws-dev

Hint: use 'tsh kube ls -v' or 'tsh kube ls --format=[json|yaml]' to list all kubernetes clusters with full details.
Hint: try selecting the kubernetes cluster with a more specific name (ex: tsh kube login test).
Hint: try selecting the kubernetes cluster with additional --labels or --query predicate.

Related issue:

• Add ability to differentiate auto-discovered databases #22438

[smallinsky]

I have tested with this with GCP audto-discover and it works fine. But wonder if we can delay backporting to v13 resource name collision CLI changes after we will finish v13 test plan regression.

@GavinFrazar
Additionally we have also the tsh proxy kube command that needs to be extended.

[public-teleport-github-review-bot]

@GavinFrazar See the table below for backport results.

Branch	Result
branch/v13	Failed