Rightly or wrongly, I assume that what is meant is that identifiers should not use the CV to enable them to identify species they are not personally familiar with.
If that’s the case, I partly agree - too many observers will auto-agree and you end up with an observation at RG based on no actual ‘identifier knowledge’. But I have used the CV, particularly with older observations, to help work out the right direction to move high-level IDs, and I don’t see a problem with that if used with discretion. I don’t go to species unless I have actual knowledge, which avoids the ‘uninformed RG’ issue.