That iLIGO golden PD circuit exhibited the transimpedance of ~2kOhm at 33MHz at the 40m.
Considering the gain of 10 in the preamplifier, the  resonant circuit shows the equivalent resistance of ~200Ohm at the resonant freq of ~30MHz.
This becomes the source impedance for the preamp. 

You can put the parameters in the formula in P18 of LMH6624 data sheet (standard noise calculation), you basically get ~2nV/rtHz input equivalent noise, which corresponds to ~11pA/rtHz current noise reffered to the photocurrent. The noise levels are n(MAX4107)<n(LMH6624)<n(OPA847), but not much change.
The shot noise intercept DC currents for them are 0.38, 0.40, 0.44mA, respectively.

This means that MAX is the best although it is a discontinued chip.
LMH is slow w.r.t. the slew rate and not that low noise compared to MAX.
OPA is fast and slightly more noisy.
MAX is known to have soft saturation, but it might also be true that all of them have this issue.

Note that they tends to be unstable. Particularly, LMH and OPA are not stable at the gain of +10. You need to increase the gain to +15 or even +20 to mae it stable. This equivalently lower the slew rate (--- should we call it "input equivalent slew rate"???). So again MAX seems a little better.