I'm afraid I disagree with most of that. I have an APS-C DSLR (a D5000) and an MFT (GF2) and the size difference in lenses is significant. The 20mm pancake is MUCH smaller than the 35mm prime I had on the Nikon - probably 1/5 the size. The MFT 14-140 is much smaller and lighter than the Nikon 18-200. I'm not a fan of 3x zooms, but even the non-collapsing MFT 14-42 is shorter than the equivalent APS-C 18-55 (probably about 2/3 length and weight).
So, if the ONLY lens you own is a 3x kit zoom, then maybe the lens size difference is negligible. Ditto on body size as the NEXs are pretty small for APS-C. But as soon as you start adding primes or telezooms to your kit bag, the difference in size and weight is significant.
The difference in IQ between APS-C and MFT is also becoming neglible. The old 12MP MFT sensor suffered from poor DR and noisy high ISO compared to APS-C, but the new Sony sensor in the EM-5 seems capable of rivalling the best crop sensors.
AF is a mixed bag - for static objects, MFT contrast AF outperforms the phase detect on most consumer DSLRs, but isn't as quick or reliable for tracking moving objects. But I understand that Sony are already working on on-chip PDAF for their NEX system, so it should be possible to add to the MFT chips they make. Nikon already have it on the Nikon 1. So this advantage of APS-C DSLR (for some users in certain conditions) over MFT and mirrorless APS-C will soon cease to be significant too.
As you say, the target market for mirrorless is upgrades from P&S. Size and weight will be a significant factor for such users, and MFT has got the jump on APS-C mirrorless there. Even if Sony and Canon can shrink their bodies to MFT size (Sony are very close), the lenses will always be bigger. And that becomes more pronounced with zooms (which appeal more to P&S upgraders than primes).
I actually see the future being smaller sensors. Shallow DoF is only of interest to keen photographers, and all the other benefits of larger sensors can be overcome with improving sensor technology. Ultimately bodies can only be so small to accommodate human hands, but smaller sensors mean smaller, lighter lenses and more in-body room for other features (flash, GPS, IBIS, wireless...). Or how about a collapsible body?....