Exactly what makes you think that the voters in November will split exactly 50% between republicans and democrats. It has never happened and won't this year, either. Trump can and will win democrat votes, but more democrats will come to the polls than republicans. That is the easiest stat to predict for this November.
Do you mean Presidential elections? Or, are you including off-year elections?
In 2006 and 2008, there were more Democrats.
In 2004 and 2010, party affiliation was equal.
But, in 2006 was the last off-year election while Bush was President, and opposition to the Iraq war finally came to a head. And, 2008 was the year Obama was elected.
I couldn't find more recent results for 2012 and 2014, but I would expect similar results: In 2012, there were probably more Democrats, and in 2014 it was probably about even.
Overall, there tends to be more Democrats in Presidential elections than there are in off-year elections. But, the "enthusiasm" factor benefited Obama -- along with the mediocre Republican candidates.
This year, I think Trump will be the net beneficiary of enthusiasm -- if for no other reason than because some Republicans will vote for the first time since Reagan. The "woman" factor will help Clinton, but not enough to offset the "disgust" factor.
I think the poll should still be oversampling Democrats, but +8 is unreasonable.
The last two presidential elections, perhaps...but the Messiah is not on the ballot this year.