By one of those stange paradoxes, applying huge amounts of energy to a tiny area in a large body of ice doesn't lead to much localised melting. Instead, the over-supply of energy blasts a crater. Much of the ejected debris is shocked to the point where it melts or vaporises.

The surrounding material remains more-or less frozen and capable of sustaining a large crater. Look at the icy moons of Jupiter and Saturn - they're generally cratered to hell unless they experienced a global or local remelting event. Even then, at small-scale they're covered with craters.

So I don't think you can use the "it would melt" argument against craters in relict ice on Mars. (There are some impressive craters in the ice of the south pole of Mars, too). Instead, arguments are better directed at to how to preserve many metres of relief and textural differences between dust-covered ice floes and dust-covered areas between floes that are still ioce-floored. - All this on a planet which is not generally known for blanketing areas with dust to the point that craters are totally obscured - if that were the case, the floes would be buried too.

I have no problems with the surfaces in question being quite young (but I doubt the 5 million year figure). I do have problems with it being water/ice. Lava is a much simpler explanation of what we see here, imho...