Now, this poll looks fine.
Looks like Clinton got a bounce that was 1-2% higher than Trump's (she was up by 3-4% before the RNC).
I'm in general agreement with you, but don't convention bumps take several days to totally manifest themselves in the polls?
That is one of the more neutral projections (Nate Silver is meticulous about that), one that says that the effects of Convention bounces are directly equal. Such has a basis in historical trends in other elections. I can say that the most conservative (statistically if not politically) assessment of the effects of the Party Conventions is that they were equal. Thus a return to where things were before the Conventions puts Hillary Clinton back in the lead. Thus the net effect of the Conventions is zero in effect on the vote -- but they devour time.
Time is precious in a Presidential race. At this stage the calendar is the arch-enemy of the lagger in the polls. This is with a zero net effect of opposing campaigning and advertising.
Donald Trump was behind Hillary Clinton going into the Republican National Convention. Should he be behind by a similar margin in late August as he was in late-middle July, then he has lost about 35 days of opportunity to catch up with about 70 days to go.
I have yet to see what a 5% advantage looks like with post-Convention polls in the states. Of course it is possible to get 50% of the vote and lose the election (see 2000) -- but not likely. With a 5% advantage, Hillary Clinton has all of the states that have never gone for a Republican nominee for President since 1988 locked down tight. That is 243 electoral votes right there. She definitely wins New Mexico, which is now strong D. That's another 5, putting her at 248.
Even without Iowa, Nevada, and New Hampshire, she wins outright if she wins Florida. I find it hard to see how she loses Iowa while winning Wisconsin decisively (the states are very similar in demographics and electoral history, so with Iowa or Nevada plus a Blue Firewall that contains New Mexico (254 electoral votes) she wins with Ohio as well. Take both of Iowa and Nevada or one if those two or New Hampshire and Virginia also wins the Presidential election for Hillary Clinton.
... As if you have not noticed, I have restarted the binary matchups for Clinton vs. Trump with two exceptions, those acknowledging that Hillary Clinton has absolutely no chance to win either Alabama or Oklahoma. I can think of no way to restart the three-way races between Clinton, Johnson, and Trump. Clinton up 5 in a binary matchup with 50% suggests 2012-style results. But even with that I have a caveat: it's hard to imagine two different Presidential nominees getting nearly-identical results on an electoral map unless one has 49-state blowouts or something like that. (Minnesota was the second-worst state and Massachusetts was the only state that Nixon lost in 1972; invert that for Reagan in 1984 and recognize that both lost the District of Columbia in those near-sweeps in 1984... Duh-uh!)
Partisan hack as I am, I am wise to let the polls do most of the talking. but I have my own gut feelings. The Conventions are where Parties ideally offer themselves at their most optimistic and where they try to solve their problems (which should be self-evident). Donald Trump expressed well what he believed in. Whether Americans like that is yet to be shown. Democrats went into their Convention with many unsolved problems. If they have solved those problems, then the polling for Hillary Clinton can get even better. They have plenty of material for negative ads against the Trump campaign, some supplied by Donald Trump and his surrogates. Even so, I iterate:
I am wise to let the polls do most of the talking. 
