Dear Mario and Ben,
Once again I'd like to add my two cents to Ben's questions, especially #1. I'm not a historian, but I think that if you examine the history of dark matter within the physics community, it's not seriously considered until the 1980s when Vera Rubin et al presented the conclusive results of >10 yesrs of observation indicating that spiral galaxies do not comply with Keplerian rotation curves. As I understand, cosmologists attempting to model galaxy evolution (apparently also using two-body evaluation methods, since their models always 'flew apart') at that time found solution for their problem in the 'dark' 'fudge factor'. Given enough invisible peripheral or spherically symmetrical mass, two-body gravitational evaluation methods work fine for any distribution of 'ordinary matter'. Following the erroneous determination that dark matter was necessary to explain the development of spiral galaxies, cosmologists began to estimate universal dark matter requirements based on estimates of galactic requirements.
If this summary is reasonably correct, the foundation upon which the house of dark matter cards rests is the expectation that galactic disk object should independently rotate around some collective central mass. I hope that assumption has been falsified.
As for gravitational lensing produced by large scale compound objects, that's a more complex issue. As I understand, the estimated mass of individual galaxies has often based on luminosity or their dynamic velocities within clusters. However, unlike main sequence stars, galactic luminosity is not directly a product of mass. The gravitational interactions of galaxies within clusters is complicated by their interaction with the massive intracluster medium, thought to contain 2/3 of total cluster mass from 'ordinary' matter. I suspect these interactions are not properly represented in galaxy cluster mass evaluations.
Weak gravitational lensing analysis is a very complex statistical process of evaluating minute optical distortions in thousands of background galaxies. From this evaluation the mass necessary to produce the identified gravitational effects is determined.
There are other methods used to confirm those determinations using CMB signals, etc. that I cannot explain. The point is that the evaluation of large scale weak gravitational lensing effects produced by compound masses is an exceedingly difficult set of processes that are each subject to many potential errors.
Personally, I suspect that the identified lensing effects are often considered to be the product of only a single, large scale curvature of spacetime imparted by the collective mass of all included lensing masses. I think that the actual lens effects produced by a large scale aggregation of discrete masses would more closely represent a compound eye, with a global lens structure and embedded discrete lens structures, in galaxy clusters for example, produced by the ICM and individual galaxies. I suspect that these compound lens effects exceed those expected from the collective mass of the entire structure.
I hope these observations help, but I'm really not capable of documenting credible sources, etc. So I wholeheartedly agree with Mario's assessment - I focus on the galactic rotation origins only because that seems so obvious to me that I keep thinking that it can't be dismissed forever! I don't know what the cosmologists would ever do with their models if dark matter were ever rejected, but I'm certain they and others will naturally resist that conclusion.
Sincerely, Jim