Finding (2) is that essentially all nominal model scenarios end up *way* too wet. Rocky planets in the "liquid water habitable zone" of their M stars all receive 1-2 orders of magnitude more water than the Earth does. This is not good. This is too much oxidation, way too much. These planets are big giant comet, iceballs, water worlds, whatever you want to call them. Our simulations only produce compositionally "Earth-like" planets (water/rock ratio) if planetesimals dehydrate substantially.
Magma ocean losses are limited to a few tens of Earth oceans. This is not enough to balance the massive accretion of water just from late impact bombardment alone. This suggests that M dwarf rocky planets should have no problem creating an "atmosphere", but actually they are not rocky, they are massive comets. Only outgassing from radioactive decay in early-formed planetesimals can dehydrate planetesimals strong enough that the resulting planets are in an Earth-like water regime.