Block time is only designed to average every 10 minutes.
Block time is designed to average 240 seconds or 4 minutes, not 10 minutes.
Apologies -- you are correct, it is idealized to 4 minutes. The rest of my logic still holds, however. I do remember a few block times of around an hour due to difficulty spikes back when original Sumo was still on CN -- ASICs would join, snap up some blocks, then leave. This lead to the difficulty spiking, and with the now-reduced hasrate, it took a long time for the rest of the network to find the next 1-2 blocks (at which point the difficulty would correct). This "situation" lead to the fork to CN-Heavy. +10 minutes, tho not ideal, isn't near as bad as +60 minutes.
You're on the right track, the fork from CN to CN-Heavy was part of the solution...it was the part to brick the ASICs.
Yet, even with ASIC out of the picture it was still possible to network hop if you had a large GPU farm and could do as you mention, join and sweep up some easy diff blocks then leave once the network diff adjusted leaving the rest of the network to hash away to solve a way to hard of a block.
Well the Difficulty Adjustment Algorithm (DAA) was a SUMO customization of Zawy12's original work, the customization had a flaw which allowed the SUMO chain to be attacked for many easy blocks. This is why the SUMO chain is now so many blocks ahead of RYO's. Psychocrypt and Fireice collaborated with Zawy12 in refining some security issues with LWMA which he backported, and since 'Free Radical' v0.2.0 Release we've been on that DAA.
As Fireice calls it, "rocking" of the network is still possible (harder to accomplish the larger the overall network hash) but the security flaws have been resolved.