A 10-core die would mean you can get a 6-core part with 40% of core's on the die failing or 20% with an 8-core failing. If 14nm's yields are still far from perfect(They started in a pretty shocking place so even after all this time I'd say it's possible) a 10-core die would make 6 and 8 core parts more abundant. The "halo" price premium would get moved to the 10-core part allowing the 6 and 8 core parts pricing to come a little closer to reality even in the face of high demand. This is the only practical way to progress with pricing & performance without a notable node jump.
It doesn't look like we're going to get 10nm dual core mobile parts until sometime deep into next year(Technically 15W dual-core 10nm parts with no graphics have already launched but you can't really class it as a mobile part given it needs a dGPU, likely because the GPU portion is often as large or larger than the CPU portion and would have destroyed yields), so chances are we won't be seeing anything larger than 4-cores on 10nm in 2019. This basically means a mobile only launch followed by a low-end desktop launch using the same dies, as they had to do with 14nm due to its similarly troublesome birth, where the 10nm-Broadwell never made it to desktop(Instead the desktop market got another 22nm refresh with Haswell's refresh). They can't launch 10-series mobile & low-end desktop parts without some high end desktop parts to partner, which I assume is where this 10-core will come in.
