I think, ultimately, it's the wrong course of action to take. One of the points people have mentioned in this thread is the feeling of discomfort/vulnerability/distraction students feel while learning alongside members of the opposite gender, that they do better when segregated--and that this relates to why women are less inclined to take physics at the A-level, because of "...initial male bias in physics lessons...".
To say nothing of the broader potential social implications of this as a way to educate a population in a Western country, it does nothing to solve foundational problems (like symptom prevention instead of cure). If they made it so females could pursue engineering without A-level physics, it would just create a power imbalance among some of the males and females in those university classes where knowledge of the relevant physics is recommended. Females in those classes that aren't especially geared towards STEM (but decided to pursue it because it became easier to access and because of the potential for lucrative future occupations) would probably begin to turn to some of their male peers for help, since presumably the males were forced to take the relevant A-levels while most of the females in the class presumably wouldn't have (or most wouldn't have independently developed the requisite knowledge).
I have many times asked peers of mine to help me understand some difficult concepts in physics and computer science, and since they were my peers, this always made me feel a slight bit less, like I wasn't as capable or as smart as them and had to rely on them for some detailed, brief tutoring. Now obviously there's no problem with asking for help, but literally ensuring a social dynamic of this kind on a broad scale, even if it is only until (or if) those females catch up to the males? Imagine for a moment that females aren't, *on average*, as predisposed to STEM-related subject matter as men are (as corroborated by mainstream findings in evolutionary psychology). If that is indeed the case, I think something like this would only breed more resentment or frustration among many of the women that decided to go into engineering without that earlier preparation. They might even suspect their is an "initial male bias" in many of the engineering classes that generally require the knowledge of A-level physics.
Also, we're talking about young people in university trying to get laid, and statistically speaking, there are going to be unethical actors to varying levels of behavior everywhere, even in the UK. Does anybody else see the potential problem here if there are some percentage of females who haven't prepared as much as they should have because they aren't required to, and all the rest of the males in the class have? Many of them will feel desperate to maintain grades and will struggle to keep up, and I'm sure there are going to be some males who will take advantage of some of those less prepared peers that they encounter in their classes. I can easily imagine a male student offering to "tutor" a less prepared female peer, with an ulterior motive in mind.
I'm not saying these potential problems can't be mitigated (for example, by requiring female students that didn't do A-level physics to take some prerequisite courses that will prepare them adequately with respect to their male peers), but ultimately it seems to me that allowing females to pursue engineering without the benefit of that prerequisite physics knowledge (and without completely changing the pedagogical approach to teaching engineering classes to account for females with less foundational knowledge) is one way to pave the path to hell with good intentions. I think it's important to get more women into STEM, but I don't think this is the wisest approach.
