This is not (yet) a solution, but a step towards it.
Consider a generating function
$$F(x,y) := \sum_{n=0}^\infty \sum_{k=0}^n f(n+1,k+1) x^n y^k.$$
The recurrence for $f(n,k)$ implies the following PDE:
\begin{split}
& x \,y^{3} \left(y-1 \right) \frac{\partial^{2}}{\partial y^{2}}F \! \left(x , y\right)
+x \left(x-1 \right) \frac{\partial^{2}}{\partial x^{2}}F \! \left(x , y\right)
+\left(x y -1\right) x y \frac{\partial^{2}}{\partial x \partial y}F \! \left(x , y\right) \\
+ &\left(6xy^{2}-2xy - 1\right) y \frac{\partial}{\partial y}F \! \left(x , y\right)
+ \left(2x^{2} y +3 x -2\right) \frac{\partial}{\partial x}F \! \left(x , y\right) \\
+ & \left(6xy^{2}+2xy+1\right) F \! \left(x , y\right)+\frac{3 x y -2 y -1}{(x-1)^2(xy-1)^3} = 0
\end{split}
with the boundary conditions $F(0,y)=1$ and $F(x,0)=\frac1{1-x}$.
I did not try to solve it manually, but Maple does solve it (more specifically, it solves the aforementioned PDE times $(x-1)^2(xy-1)^3$) in general, but cannot take into account the boundary conditions. This gives a hope that the PDE is soluble for the given boundary conditions as well.
Having an explicit formula for $F(x,y)$ will likely provide one for $f(n,\lceil n/2\rceil)$, which then can greatly help in establishing the inequality in question.