Enable tolerance to timed out poll requests

[raliev12]

Problem description (if applicable)
There is quite a common situation when apply process interrupts due to timed out poll request, i.e.:

[airshipctl] 2021/06/01 21:09:28 opendev.org/airship/airshipctl/pkg/events/processor.go:84: Received error when applying resources to kubernetes: polling for status failed: etcdserver: request timed out Error events received on channel, errors are: [polling for status failed: etcdserver: request timed out]

However it's not necessary to completely interrupt apply process due to this error, we can try to poll resource's status once again before giving up. In most cases, the next request completes without an error. We can observe such errors in our CI jobs:

https://jenkins.nc.opensource.att.com/job/development/job/Airshipctl/8927/consoleFull
https://zuul.opendev.org/t/openstack/build/336b136065ef4daa95bcc1fc15f1ca5c

Proposed change
We can filter specific errors we want, and therefore apply process won't be interrupted unless we accumulate threshold of these errors (like timed out). To do this, we need to make some changes in our poller implementation and actually use it (right now we use the default one).

Potential impacts
None.

[raliev12]

Related PS - https://review.opendev.org/c/airship/airshipctl/+/794087