My point of view says that when we say this it a circle , then we defined a region (or our shape is definite ) Whereas in case of total distribution curves of electron , they decays but do not became zero ,this means electron is spread through out the space. Logically you can't define its shape until it is not definite.

In case of boundary surfaces we are taking an approximation that we are neglecting some of the fraction of orbital so that it become definite.

and if I talk about whatever you said , then I have a counter - reasons to prove why orbitals do not have shape.

ψ = f(R,θ,Φ) or Ψ=f(x,y,z) , this equation contains 4 variable therefore if you actually want to visualize a continuous plot of this equation you will need 4 dimensional graph. And you can't visualize any 4-D shape.

and if you are plot it by shading then it will not be continuous . It is equivalent to draw these graphs

i.e x^{2}+y^{2}+z^{2}=8 in 2-D by shading

or y=2x in 1-D by shading.

you have to take approximation(by considering its boundary surface) or draw a quantized graph (by shading it) .