Quoting Mike here on the need for this from a supportability perspective:
Upon reviewing our code I realized that we're deleting persistent volumes when we delete pods. This means that in some cases we're deleting logs that may contain valuable debugging information before users can get a chance to collect logs from them. Without also fixing this issue we wouldn't get much benefit from adding a persistent volume to the logs directory for clusters with ephemeral storage because the logs would likely get cleaned up after a failure anyways.
We can disable volume deletion by default and add a flag to enable it if user wants it.