There is no current non-invasive way to measure blood glucose in the medical industry - if there was, every diabetic would already be using it and we'd be using it in hospitals. We just aren't. Most of the devices floating about now are hit & miss, not very accurate and require calibrating with an actual blood sample.
I'm not saying it's impossible that Apple might come up with it, but it's pretty damn close to impossible. Plus, I doubt Apple is going to attempt to go down the road of making the Apple Watch a 'medical grade' device - slightly more hoops to jump over there (FDA in the US for a start)... for good reason. If it's not accurate, it will cost lives.
People who think blood glucose can be detected the same way the pulse is detected on the watch (using differential absorption of light), clearly have no understanding of physiology, physics or biochemistry... current monitors make a lot of assumptions.
And for non-diabetics, what the hell are you going to do with the data - the signal to noise ratio shrinks even further, yet we think we're all 'well informed'. The Dunning-Kruger effect on display.
Quite a few years ago there was a glass ampule type of thing that could be injected/inserted under the skin that could do the job of estimating blood glucose. I have a friend who has injected a similar shape/size RFID tag under the skin of his hand. He uses his hand for many RFID key-door things in his life. (In fact your pet cat probably has the same type of RFID tag injected - mine does).
For a patient diagnosed with diabetes, maybe it is not such a big invasive deal to have to inject some sort of glass capsule under the skin, if it could really help measure blood sugar.
Anyway - technology advances, and I would bet that less invasive methods will be developed. There is a need for it.
It might seem difficult or "close to impossible" as you say to non-invasively measure blood glucose. But once you see how the magic trick of doing that is done, you might say, "well yeah, I guess you could do it that way".
Just look at helicopters! Did you know that for 20+ years it was thought that many helicopter designs were inherently unstable and not worthwhile? I know some of the guys that had to work to find the solutions despite the aerodynamics textbooks telling him it was "damn close to impossible". The advances came from the TOY industry - NOT from Darpa grants or Lockheed Skunkworks.
These days we have many very cool helicopter designs! People look at them and say "yeah, I guess you could do it that way", but they didn't do anything at all - it took someone trying to accomplish "the damn near impossible".
All it takes is a fragment of a "success" to spark the imagination of the next inventor. If he can see the light because of some seemingly small advancement, he will take it to the next level and eventually we get a breakthrough!
I'm positive that the problem (of non-invasively measuring blood glucose) will be solved one way or another.