Uncertainty from rounding of constants.

This is about the rounding of mathematical constants like the number pi. (slides D-SI v1.2 page 22, original comment from NPL)

The pi example is potentially confusing. The uncertainty is a numerical uncertainty (that arises from rounding) rather than a measurement uncertainty, and perhaps needs to be treated differently to a measurement uncertainty. Using a value of pi (3.14) that is rounded to 2 decimal places leads to the value of numerical uncertainty you quote, but then providing the value of pi to a number of decimal places (3) that reflects the precision of the uncertainty value gives an estimate of pi of 3.140 when one would expect to see a value of 3.142. Perhaps some discussion is required?