According to Mathematica, the determinant of the Hessian matrix of $(x,y) \mapsto \sqrt{\frac{1}{1+x}+\frac{1}{1+y}}$ is
$$ \frac{3}{4 (x+1)^2 (y+1)^2 (x+y+2)}$$
Its trace is
$$\frac{3 x^4+4 x^3 (y+4)+6 x^2 (2 y+5)+4 x \left(y^3+3 y^2+6 y+7\right)+3 y^4+16 y^3+30 y^2+28 y+14}{4 (x+1)^3 (y+1)^3 (x+y+2) \sqrt{\frac{x+y+2}{x y+x+y+1}}}$$
They are positive for $x,y \in [0,1]$, showing that the function is convex. Therefore
$$(a,b,c) \mapsto \sum \sqrt{\frac{1}{1+a}+\frac{1}{1+b}}$$
is also convex.
The key now is understanding that moving variables apart increases the objective function. Obviously $a,b,c \in [0,1]$. Suppose that $0<a,b,c<1$ and $a\leq b$. Replacing $a,b$ with $a-\epsilon, b+\epsilon$ gives
$$ (a-\epsilon)(b+\epsilon)+(b+\epsilon)c+c(a-\epsilon) = 1-\epsilon^2+\epsilon(b-a).$$
For $\epsilon>0$ small enough the value of the constraint function is increased so to preserve equality we must choose $c_\epsilon<c$ such that
$$ (a-\epsilon)(b+\epsilon)+(b+\epsilon)c_\epsilon+c_\epsilon(a-\epsilon) = 1$$
Convexity shows that moving $a,b$ apart increases the objective function. Decreasing the variable $c$ further increases the objective function. Thus if all $a,b,c$ are strictly between $0$ and $1$ they do not maximize the function.
Thus, at least one of $a,b,c$ belongs to $\{0,1\}$. Continuing the casework will lead to the solution.
This is not an elegant solution, but hopefully it will lead to one.