Better Container termination status when failing TaskRun for Image pull errors

[JeromeJu]

From #4921 , we introduced the mechanism for failing the TaskRun when any of the step fails with the image pull error: ImagePullBackOff. It also then marks all other steps in the Task as failed with the same reason "TaskRunImagePullFailed". This issue proposes the improvements of the error handling:

• Mark other steps terminated as a new pod status reason, instead of marking them all by a Taskrun reason, to clarify that it has not been executed: something like CanceledPreviousStepImagePullError
• We could add specific step, image being pulled to the &corev1.ContainerStateTerminated at failTaskRun.
• Cleanups on the existing error messages and controller logs.

We shall also explore the extensibility of step/container to examine how we could distinguish TaskRun reason and Step reason more precisely for a better user debug experience. For example, better clarifications of step failures for image pull errors.

/kind bug
/kind feature