Currently, when the ILM downample step is being retried, the same target index is used.
This can cause the subsequent downsample API invocation to index rolled up data into
shards of the target index that already exists and while the previous downsample API
invocation is still partially running (and also rolling up data into the same target shard).

Note that, the downsample step may fail in case a cluster is being restarted in a rolling 
manner (for example for an upgrade) or when the elected master node fails (the downsample 
action is coordinated from the elected master node).

This PR modfies the ILM DownsampleAction so that when DownsampleStep fails, it will retry
by going performing the following steps
  1. Cleanup existing target index, 
  2. Generate a new index name for the target index
  3. Downsample using the new target index name.

Note 1: This change may leave some garbage indices that we must find another way
to cleanup. However, the downsample process will become more resilient.

Note 2: A similar approach is used by the searchable_snapshot ILM action

Closes elastic#93580