I am well aware that it is industry standard, but when you get to this pronounced of a shift, it becomes a question whether the entire sample should be thrown out (aka the 1 of 20 polls that is unreliable as a matter of science). I don't think I've seen a poll that when accounting for demographics shifted the D/R/I by 8% before.
What people like you don't understand is: PARTY ID DOESN'T MATTER THAT MUCH.
Party ID is merely a state of mind question, not a concrete demographic. Party registration is, but that's rarely asked in polls.
Didn't you learn anything from 2012? Unskewing by Party ID has never made any sense.