I like how during our tests I tried to explicitly point out things that I didn't want Moses to misinterpret yet he has still done it.
So to summarize the tests we have done I've concluded that:
in a straight up fight with little micro Bricks win very heavily vs ML
In a fight where player does micro his ML well but not perfectly and he also can approach only from 1 side (99% of situations in a real game) Bricks win vs ML with around 8 bricks remaining.
With near perfect ML micro and ML approaching from 1 side it is possible for the ML to win but it's really really close and it comes down to micro and pathfinding of bricks.
If ML is allowed to run circles around bricks and poke from every side and the player micro's ML well then ML wins.
So from that we can see that in ~90% of situations ML looses to bricks pretty badly, when the player manages to micro near perfectly the ML draws or wins, and in the 0.1% of weird situations where ML is running circles around enemy forces it wins easily.
Few more things to consider: The micro that ML requires is basically going into range of bricks while turning and starting running away at the same time so the bricks can easily disengage and thus buy the defending player a lot of time to make more units, finish his own experimental, make pds air or w/e. So if you want to attack with ML, then ML looses 100% of the time making it an inefficient investment mass for mass.
I hope this cleared things up a bit.