Improve k8s_wait_for_deployment

[antgamdia]

Description of the change

This PR tries to fix the current k8s_wait_for_deployment function behavior used in CI. When the status is Terminating it just exits without considering the timeout. In a previous attempt, we added an additional fallback attempt, but it seems it's not enough.

Example: see how it's terminating, after 10 grace seconds it tries to get the status and finally it errors (with just 2 deployments left).

TEST SUITE: None
INFO  ==> Waiting for Kubeapps components to be ready (local chart)...
INFO  ==> Rollout exit code: '1'
INFO  ==> Retrying after 10s...

NAME                                                             READY   STATUS        RESTARTS   AGE
kubeapps-ci-65bbf5db54-j27zv                                     2/2     Terminating   0          21s
kubeapps-ci-internal-apprepository-controller-78cc7f5b9b-lg67z   0/1     Terminating   0          21s
kubeapps-ci-internal-assetsvc-758d68d489-4c56w                   1/1     Terminating   0          21s
kubeapps-ci-internal-dashboard-65bbc689f6-hk6st                  1/1     Terminating   0          21s
kubeapps-ci-internal-kubeops-5f7955746-9jsjr                     1/1     Terminating   0          21s
kubeapps-ci-postgresql-0                                         1/1     Terminating   0          21s

INFO  ==> Rollout exit code: '1'

NAME                                              READY   STATUS        RESTARTS   AGE
kubeapps-ci-internal-assetsvc-758d68d489-4c56w    0/1     Terminating   0          31s
kubeapps-ci-internal-dashboard-65bbc689f6-hk6st   0/1     Terminating   0          31s
CircleCI received exit code 0

Benefits

The PR #2908 will (hopefully) pass the checks again

Possible drawbacks

N/A

Applicable issues

N/A

Additional information

This approach is already being used when installing the OLM, so this change is also kind of consistent with it.