Don't look at fivethirtyeight's polls/polls-plus predictions right now. Jesus. (I know it's the post-RNC bounce, and things will look really different next Friday, but DAMN. I mean. HOW. WHY. HOW.)
It's not a post-RNC bounce. There aren't any polls covering the period past about 20 July yet. The current estimates are heavily influenced by a set of state polls from Quinnipiac which are particularly favourable to Trump. Quinnipiac is pretty reputable, they're no Rasmussen; but from what I understand, I think they might be unusually generous in their assumptions on turnout among white voters. (That said, I think it's clear the conclusion to the email investigation did hurt her.)
I'm not sure how much of a bounce Trump will get from the RNC, if any. It wasn't a complete tyre fire, but they did have the plagiarism scandal and Cruz' PUMA hatchet job. The optics and headlines aren't terrific. And the content was pretty much wall-to-wall dystopia. Conventional wisdom is that hopeful messages usually play better. Sure, the party faithful will love it (and inexplicably seemed to regard Trump's speech as positive), but will it sway voters who weren't already on board?
I think the post-conventions bounce is pretty much Hillary's to win at this point.