Remove SHOULD for advice to pre-generate KeyUpdate

[ekr]

The process of creating new packet protection keys for receiving packets could
reveal that a key update has occurred. An endpoint MAY perform this process as
part of packet processing, but this creates a timing signal that can be used by
an attacker to learn when key updates happen and thus the value of the Key Phase
bit in certain packets. Endpoints SHOULD instead defer the creation of the next
set of receive packet protection keys until some time after a key update
completes, up to three times the PTO; see {{old-keys-recv}}.

I don't believe we had consensus on this -- and there was considerable debate in YUL. It just got pulled in as part of a big PR.

This is a significantly more complicated mechanism than just generating as needed, and I haven't yet seen any real security analysis.

[martinthomson]

This text was specifically put in to address #2792. That issue was extensively debated, but I agree that it was not concluded.

[mnot]

Reopening #2792. We need to be more careful about PRs that close multiple issues...