I could write it up as a more detailed case one day, not today. But for now, note that I think we are looking for different sorts of proof. Coming from a scientific background, I'm used to making a hypothesis and then testing it by applying it to different situations. The hypothesis that is most consistent with all situations is considered "proven" until such a time as a situation arises which contradicts it.
In this case, every possible proposal (e.g. "one flesh = marriage" and "possession = marriage" and "possession + one flesh = marriage") work find in normal, ideal circumstances. If you possess a woman and are sleeping with her, we all agree you're married regardless of which of the above definitions we are using. Just because a definition works here doesn't tell us anything - all definitions work here.
It is the difficult, borderline cases where we find out whether our hypothesis actually works. And it is there that "one flesh = marriage and gives ownership of a woman" really breaks down - cases of rape for instance. These become "weird" as you have described, and some contortions are needed to shoehorn the situation into the hypothesis. To me, this suggests that the hypothesis is flawed.
While the position that I am holding - that possession of a woman actually matters - still works fine in the borderline cases. If you possess a woman, you can legitimately have sex with her, becoming one flesh. If you don't possess her you cannot legitimately have sex with her, because she's not yours - if you do have sex you still have to sort out the possession side of things. This works for rape, betrothal, prostitution - any situation you can find that is outside the "ideal norm" of a fully-formed legitimate marriage. It has clear explanatory value, and is not contradicted anywhere. To me, this suggests that the hypothesis is proven until a situation can be found that contradicts it.