Hello Manam and welcome to the forums, I like that your name is a palindrome!
Some notes:
BTN raises to 3bb, he doesn't bet. You could also say he, "opens to"
BB should almost always 3B with AQ vs. a 3bb rfi strategy from the BTN at equilibrium, so once he does this and we go to a flop where you are evaluating the strategy using a solver I'm under the impression you might be using the software incorrectly even if the output seems reasonable. Specifically, you might be inputting constraints incorrectly.
Software such as GTO+ is basically a modeling software that is finding the nash equilibrium given a variety of constraints. One of the main constraints would obviously just be the rules of the game, but other constraints which are inputted by the user would be things like: board texture, ranges for ip/oop, allowed bet sizings, stack depths, allowed raise sizings, and possibly many other constraints. If you are inputting constraints incorrectly, then the outputs you get are more or less worthless: garbage in - garbage out.
In theory the EV difference of betting basically any hand here in a BTN vs BB situation when compared to checking should be very marginal in the equilibrium output. It also should be noted that calling a hand like A7 a "bluff" on this flop is also not really a good way to think about it; it would also really depend on the sizing used by the BTN player.
Below is an image from a "simple" sim where the preflop range for the OOP player is basically an equilibrium range vs a 2.5x rfi sizing from the BTN and the OOP is forced to check and IP can either bet 30% or check. You can see below almost all actions are mixed and therefore in theory and at equilibrium within the constraints that I've inputted for the modeling software--there is no discernible EV difference between the options for ~all hands:
And here is showing the EV difference between the two actions:
You can see in the above image that the only hands that *strongly* prefer betting over checking are basically the strongest hands in IP range... sets and strong KX... they gain 0.25-2.669 more chips by betting compared to checking in this sim and with these constraints. It should also be noted to provide additional perspective that 10 chips is 1bb for this sim and the rake was 5% capped at 2bb.
It's going to be virtually impossible to make a "blunder" by betting in this node (i.e. BTN vs BB SRP on K42r) with basically any hand unless the bet sizing used was very large. If you were to plot the EV difference between different allowable sizings and check, then the EV difference between very large sizings and checking would be largest for hands that have hand strength somewhere in the "weak to middling" range... that is to say... if you were to use a very large bet sizing in a sim with a hand that has some showdown value like ace high, and more specifically strong ace high (AT-AQ), then there would be a considerable EV difference and it might be considered a blunder, but in general there really is very little you could do with this hand on the flop that would be considered a blunder.