Give it a try and let me know. I implemented the noscf keyword to get non-selfconsistent energies based on vectors from a different calculation (xc functional) but I didn't check optimizations. Try the following in your input.
dft
xc some_functional_1
end
task dft energy
dft
noscf
xc some_functional_2
end
task dft optimize
This should do an optimize based on the second_functional