To simplify discussion, I would ignore MEC when designing relic and legendary. Not everyone has MEC. I would start with a relic that is coherent. Once it's determined what it does, move on to legendary. MEC, finally, gets changed based on functional relic and legendary. More "agile" development than waterfall, as MEC just seems doomed and, thus, can do anything at this point that doesn't screw the other tokens. While could start with legendary and work down to relic, that seems far more difficult than working up from relic.
I started writing out my thought process, but it ultimately went nowhere. The problem seems both simple and hard. Simple relic is just increase damage by at least +6 while giving some other benefit. Not SC based, not messing with how sliding works, no more spells, no FA. Once an effect like FA is added into the mix, it just becomes a mess of interdependent abilities.
Hard because the benchmark for damage is really high and it's actually pretty arbitrary. Again, we aren't talking about someone with a lucky treasure draw being the target audience for relics and legendaries, the audience is the forum arguer. If all the relic did was add 9 to spell damage, besides being boring, it would not give burst damage like crits do or Sneak Attack can, so it doesn't solve the damage problem. One extra MM in a round is 22 damage, where someone suggesting the relic be +20 and the legendary be +40 spell damage is probably not going to fly. The common currency is just so messed up. That MM might only be once a combat rather than once every round, which completey changes the numbers.
I don't know. The more I think about it, the more flaws I can see in my own attempt to address the problem as what people want varies, while there are multiple dependencies to be able to set a numerical value that seem like they could change. Then, I feel the wizard pain more at every level besides the level where relics and legendaries matter because class card changes matter so much less at the relic/legendary level.