Summary:
During K8s volume resize, we do helm upgrade to recreate STS. But instead of using only the
modified volume size, we use the entire new task params in helm upgrade subtask. This is an
issue when combining volume resize with other spec changes, as it leads to pod restarts before
reaching the actual subtasks which take care of modifying the spec. This is not a regression
as before master volume feature too, we used to apply KubernetesCommandExecutor params with
taskParams() based universe details itself.

Fixed the behavior by only using new device info changes when doing helm upgrade for volume resize
case. Also made the volume resize go from zone to zone instead of all at once.

Test Plan: Verified by combining volume resize for tserver/master with resource spec changes.

Reviewers: dshubin, anabaria, anijhawan

Reviewed By: dshubin, anabaria

Differential Revision: https://phorge.dev.yugabyte.com/D44026