Note why the Redis hash tag keeps the rule name

Records the reasoning @reprazent asked for when merging labkit!346, and shortens the comment while doing it: nine lines down to four.

The hash tag could hold only the identifier characteristics, which would let Evaluator#clear delete an identifier's keys in one DEL instead of one per rule. It does not, because that would put every rule's key for one identifier on a single cluster node, so a hot IP would concentrate load rather than spread across the cluster. The per-rule DEL is the accepted cost.

Comment only, no behaviour change.

Related: !346 (merged)