I've got a simple advice for you: Forget about starting to add zeros and poles to stabilize a voltage regulator. Like you mentioned, you don't control the output, so unless you narrow down the load capacitance value and esr values to impractically small ranges you are going to find a condition where your extra poles and zeros will hurt you and make the system oscillate over process, temp or both.
In LDOs the key is to keep it simple, accept with happiness whatever zero you get from the esr in the load capacitor. It is a pretty difficult AC problem to begin with, adding more AC complexity won't help. If internally compensated, add a Miller cap and that's it. Users of LDO's can get creative, because they are only worried about their application so they can get fancy adding external components that will work for their specific case. Good luck.

Jerome,

 Sorry I can't access Xplore right now, but I think one of the two papers
below has the information you are looking for. As I remember there are
some trade-offs in doublet compensation. Large overshoot that settles
faster or less overshoot that takes a long time to settle. The trade-off
between overshoot and settling time has to do with how well the zeroes
compensate the poles. In the real world the compensation is never
perfect. The larger the mismatch, the larger the overshoot and faster
the settling.

Analysis of the settling behavior of an operational amplifier
Chuang, C.T.;
Solid-State Circuits, IEEE Journal of
Volume 17,  Issue 1,  Feb 1982 Page(s):74 - 80

R. J. Apfel and P. R. Gray, “ A fast-settling monolithic operational
amplifier using doublet compression techniques,” IEEE J. Solid-State
Circuits, vol. Sc-9, No.6, pp332-340, Dec. 1974.

Also I think that Raul is right in that for your application, you need to
design your LDO around the capacitor.

                                                            Best Regards,

                                                               Sheldon