Yes there is a bit "latest greatest" envy, but the speed bumps won't be from the microarchitecture, it will come from the process shrink. There is little generally wrong with a process shrink as it typically leads to "faster".
For those with workloads that scale by core Ivy likely means another two cores in the single package line up and another 4 cores in the dual package line up. That will be much more than 5-10% on those workloads.
We'll see. If they cap the E5 1600 cores that their current level for Ivy (at 6), it is probably even more likely will see some iGPU additions to the E5 line up when get to Haswell. If they move the 1600's to 8 max that will be not so good sign for iGPU at next step.
You have great points and if they squeeze more core in then great they are already having issues. Mine is more a question of why people that have no idea what you are talking about prefer to have Ivy and are waiting because someone else says they should because they heard it from someone else. So it must be just Intel marketing saturation that the mindshare for Ivy is so prevalent. I don't even remember i7 having this much brew-ha around it and it actually boosted higher performance tick vs. tock.