The circuit linked by albbg does the opposite of the original circuit, dividing the R feedback branch instead of the C branch, so it doesn't explain anything.
A reason for the divider may be, that the OP isn't unity gain stable, but the effect depends on the diode capacitance and OP GBW, without this information, everything is just a guess. I should add, the assumption, that the original circuit designer did something resaonable isn't more than a guess.
As Keith mentioned, it can be used as a means to set small C feedback values. But 3.6, or more exactly 3.3 pF is still a convenient capacitor size.