Time. Women live longer on average, and there is also on average an age gap between heterosex partners (males older); therefore it is more probable that the male sex partner will be deceased, not the female. The numbers for females, if anything, should therefore be higher.
Space. If the UK is a net exporter of male sexual partners, that would mean that some other country or countries would be net importers. Regardless, the worldwide number of heterosex connections must be the same for women and men. But that's not the case. In every country, with the notable exception of New Zealand, women report less sex partners than men, and the worldwide aggregate for men is still unexplainedly higher than the aggregate for women.
Some studies suggest the gap to be due to women underreporting, some studies suggest prostitutes are underrepresented in the samples, but in the end, whenever somebody wants to argue that men tend to have more sex partners than women, let them know that (a) one number is pretty useless in telling the whole story, and (b) if conclusions of a study present a mathematical paradox, they should be taken with a grain of salt.
Uh, no, not from the sets under study. You've ignored time and space. The sets under study are the set of men and set of women currently living in a particular countries. The sets from whom their past partners are drawn (even assuming purely heterosexual relationships) may include people no longer living at the time of the survey, as well as people who are alive, but not living in the country under study (either at the time of the survey, or even ever.)
If you managed to get responses from representative sample of all men and women who had ever lived (which, you know, is logistically tricky), then you could validly argue that all heterosexual partnerships must draw evenly from both groups. But you don't, so you can't.
The imbalance could just be that UK men are more likely to have had past relationships with women not currently living in the UK than UK women are to have had with men not currently living in the UK. (It could also be the kind of reporting bias you suggest, and there are certainly other sociological reasons to suspect that such reporting bias is at least partially to blame -- its just not, even without bringing into it questions of same-sex pairings, the kind of mathematical necessity you make it out to be.)