Yes! Yes! Yes!
Of course every advance deserves to be rewarded with a feature request, in this case:
is it possible to load a common copy of the C runtime library into the internal RAM of one of the cores, such that it can be referenced by core programs of the other cores?
Given the limited internal RAM and poor speed of external RAM, it would be interesting to measure the effectiveness of executing code from another core's memory - especially if by sacrificing one core it becomes possible to save 15x the internal RAM across the others. The main pitfall I can see is that SPMD programs tend to be doing the same thing at the same time, so that core's connection to the mesh is likely to become a hot-spot.
P.S. the e_port endpoint approach looks promising.