Hopefully, you don't mean to play with your measurements. This is like scientist "playing" with the data to get the result he wants.
To give you an impression, how measurement errors reflect any estimate:
I also tried using Least Squares in a "disjunctive" approach (repeated two-dimensional) which seem more in line with what my sources tell me about grading. They say, for every last length increment in mm, there's an corresponding increment in all the other dimensions in mm.
So, I constructed a "reference last" that would be perfect for an average foot shape of Sizeadvisors' members. Since an increment of half a unit in UK sizes is an increment of 4.235 mm in last length, my current estimates for the grading procedure are thus:
Foot length: 4.235 mm * 1 = 4.235 mm Ball girth: 4.235 mm * 0.54524508 = 2.309112914 mm Ball width: 4.235 mm * 0.25838468 = 1.094259120 mm Heel width: 4.235 mm * 0.18921089 = 0.801308119 mm Instep girth: 4.235 mm * 0.58475284 = 2.476428277 mm Heel Girth: 4.235 mm * 0.99456132 = 4.211967190 mm
Put differently, mis-measure your ball width by a mere Millimeter, and your shoe size estimate for the "reference last" might be off half a unit if we would only consider this unit.
To be honest, your current measurements don't reflect a high instep.
Your instep is probably best indicated by the (short) heel girth measure. When I compare your values to the expected average for someone with your foot length, yours lie 8.359mm below the expected value. This translates to almost a full size on Sizeadviors' internal "reference last".
In other words, if we'd only consider heel girth, you should be wearing a UK 5 on Sizeadvisors reference last.
Although this is not the same as C&J's 348 last in E, I still found the information useful. First, I expect every last to behave like this. Second, the average last probably reflects the average of all available lasts to some degree.