Improve Calibration Procedure - Anomaly Detection

[drspangle]

Right now, the calibration procedure relies on assumed good measurements for each of the pins of known diameter. A better approach would use repeated measures, then taking the average of these measurements. However, it's possible that erroneous values can creep into the calculations based on the precision of the sensor, and this would drastically affect the operational accuracy. The repeated measures should be combined with a statistical test to ensure that calibration measures are within a certain confidence interval of the likely "population" average, and if they aren't, they should be rejected.

This issue ties in with #4 as it will be necessary to incorporate user feedback in order for the state of the system to be apparent to the user.

The theory of operation and statistical test(s) used should be documented.