We should lower the error level for StaleShardVersion errors on the TTL Monitor because:

• We can recover from it: next time that TTL task is executed it will probably succeed: as part of the error handling we trigger a refresh of the shard filtering information.
• We have other tools in our pipeline looking for errors.