The big issue with Megrim is that it doesn't nail an opponent for previously discarded cards before it comes down. Why this is relevant is that it rules out a strategy of using lots of D7s until you find Megrim and killing them that way. Jar decks could pull it off since the discarding happened at the end of the turn.
A correlary to this problem is that fact that Megrim does nothing on its own, so it is important to make the rest of the cards in the deck pull their own weight in order to avoid the problem of a hand full of one piece and nothing to do until you find the other. This means that you want to play good discard spells, but the problem is that all of the best/most efficient discard spells (Duress, Cabal Therapy, Thoughtseize, hell even Hymn to Tourach) work best when you play them early.
Check the "Suicide Black" list from the top 8 of last year's Vintage Worlds, played by Vincent Forino. That deck ran, iirc, 4 Duress, 4 Therapy, 2 Extirpate all maindeck. This is a potent disruption suite capable of plays like removing all of an opponent's Force of Wills from the game on the first turn. But Megrim would be useless in that deck because those spells will all be spent by the time 3 mana is even available (or else you will use a Dark Ritual to play Megrim early, giving the opponent time to play more cards before getting hit by discard).
This leaves Draw7s, which Hi-Val, as quoted above, explains to be equally poor. Oscar Tan went so far as to write an article once about cards like this in which he actually
named the problem where an engine card is useful at a different point in the game from the rest of its engine as the "Megrim Problem" and gave a lot of strong evidence why, regardless of meta, such cards can never really be effective (barring something like 4xMemory Jar, which is not a testament to Megrim's power so much as a corner case involving a bizarre quirk of a specific overpowered card).
