refactor fatal and error in unit tests

We have a mix of t.Fatalf and t.Errorf in our existing unit tests. It's most of the times user's choice to decide which one to implement. There are no specific guidelines as part a tekton pipeline development process.

As part of this issue:

• Task 1: Define a specific guideline around t.Fatalf and t.Errorf and document them as part of our unit test best practices.

• Task 2: Revisit our code base to make sure we are aligned and following with the guidelines documented in Task 1.

See relevant comment - https://github.com/tektoncd/pipeline/pull/6596#discussion_r1188598706