Having worked in the power generation industry and DC motors/drives for Stal Laval, ASEA & ABB for 38 years I have seen quite a few permanent magnet and residual magnetism machines. You are correct Nev in that mild steel is pretty poor at retaining magnetism so the actual residual magnetism can be pretty haphazard, has to do with material, homogenous structure, air gaps, etc...
I can say that Paul's method has great merit in that residual magnetism has to do, to a certain extent, with the "memory" of the material and swapping the field around is an easy and effective method of taking advantage of that memory. Use of a meter to double check polarity is needed of course.
I will add here that the magneto magnetism will have a very minor influence on the magnetism in other parts as the distance between magnetized pieces is quite large ( in magnetic field terms).
I have never had one of these Ereg units apart but the design and components in use would generally require that there is enough voltage produced to exceed the minimum break-over voltage which you do not have on the mechanical regulators as there is just a set of points (dead short) initially in the circuit. Therefore the experience of revving the engine to make it start charging is just a product of having to get the resultant voltage from low residual magnetism to overcome the break-over point on an Ereg or even to excite the field coils in some cases on a mechanical reg.
Hope this is of some help. Frank
'51 SQ4, '37 VH500 Red Hunter, '56 SQ4 (basket case), '49 Sq4 (basket case), '48 A7 BSA Twin, too many other BSA's, Norton's, Triumph's, Ducati's and Japanese to list. Just crazed is all my problem is.