Probably not. You only need a few seconds of propellant for landing: a mostly-empty stage weighs much less than a full one, and even one engine is enough to slow it down real fast. I've done some really rough guess-numbers before, landing on 5-10% of a full propellant load needed for landing.

It's probably not even as bad as that, since they certainly fly with some performance reserve already, at least several percent, and most payloads are probably not max-weight. The landing propellant in many cases may be entirely free, mass-budget-wise. They do add some additional mass for the legs and fins, in addition to any extra propellant they need, so it's not free, but the first stage is not as performance-sensitive as higher stages (see: rocket equation), so they can get away with it. I don't see 'em getting second stages back any time soon, though, and I think they've given up on that, at least for F9.

WRT engines, the efficiency based on air density is mostly a function of the nozzle, and the upper-stage Merlins have a nozzle extension to deal with this. Higher-efficiency fuels (i.e LH2) are historically preferred for upper stages, since they are more performance-sensitive, so SpaceX is losing some performance there. But LH2 is super-cryogenic and a beast to work with, and an entirely different engine. SpaceX's thing is that they are willing to sacrifice maximum efficiency for reasonable cost, so they decided to make that tradeoff. It seems like a good plan so far.