Thats not really statistically significant.
Especially when polling companies have a tendency to under poll Trump
(remember Seltzer??)
Neither shift is statistically significant and I am sure RCP would agree with you.
RCP is showing a largely flat trend.
Others showing a slight decline.
Not a lot to see here.
The only thing that seems very consistent is anyone saying there is strong, surging support for Trump and his policies is lying.
As for the polling misses on Trump, they've been quite specifically tied to
voting.
One thing Trump has consistently shown is that voters who are not normally likely voters will go an vote for Trump, which means models that try to take into account if someone is a likely voter underestimate his strength.