I believe a gluccometer model has to be within 20% of the lab results to be certified for sale in the USA and Australia, but most have tighter tolerances. But I think an individual meter is very reliable for point to point comparisons. So if your meter said your BG was 100 then it could be from 80 to 120, but if you tested again in an hour and it showed 90 then you can be pretty sure your glucose had moved 10% lower.
One trick for reducing inaccuracies caused by the collection method picking up glucose on the skin, is to express a larger droplet of blood and put the collection surface of the strip in the middle of the droplet where there will be less contact with the skin surface.
I’m interested in your experiment but I suspect you won’t see as much difference as the purpose of supplementing with fat to support fasting is to add energy to make up for the fact that body fat is capping out. Protein is also energy but not as dense as fat. At 40g of fat you are proving 360 kCal/day of energy. That’s roughly the same energy as eating 90g of protein (not sure what your lean mass is so don’t know the exact amount you’ll be eating).
I suspect the biggest difference will be internal where the fat supplemented fast should provoke a maximum rate of autophagy after 3 days. The protein supplemented one probably wouldn’t.