fix: Gate nodes associated with a machine based on resolved .status.providerID

[jonathan-innis]

Fixes #

Description

Prior to this change, in the release v0.28.0-rc.1, if a Machine failed to launch, the .status.providerID of the machine could possibly not be resolved. If this was the case, the Machine would eventually get deleted due to a timeout on registration, causing any nodes associated with the machine to also be removed.

In this case, we were not guarding against the case where the providerID has never been resolved, meaning that we were deleting all nodes that had an empty providerID, meaning that Karpenter would delete nodes that weren't managed by it.

This change guards this case so that the next release v0.28.0-rc.2 will not terminate these nodes when machines with empty provider ids are removed.

How was this change tested?

• make presubmit
• Manual Testing of Machine failure deletion

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[coveralls]

Pull Request Test Coverage Report for Build 5063074931

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 22 unchanged lines in 4 files lost coverage.
• Overall coverage decreased (-0.02%) to 81.271%

---

Files with Coverage Reduction	New Missed Lines	%
pkg/operator/controller/typed.go	2	93.55%
pkg/scheduling/requirements.go	2	98.5%
pkg/controllers/node/controller.go	7	70.59%
pkg/utils/machine/machine.go	11	39.37%

Totals
Change from base Build 4997439332:	-0.02%
Covered Lines:	6904
Relevant Lines:	8495

---

💛 - Coveralls