Make the data stream rollover more flexible #70905
Description
Replace the validation that checks for conflicting indices in a data stream's backing index namespace with a more flexible data stream rollover.
If a new write index that is created during a data stream rollover has a name that already exists in the cluster then the data stream rollover should increment the data stream's generation and perform the rollover again. This can be retried in rare case that there is another name conflict.
The current validation logic has caused problems during a cluster upgrade and currently prevents performing restoring a snapshot with data streams in a cluster that already contains these data streams.