feat: new method for user data calibration and collection

One important factor to detect gestures is preliminary motion since every gesture has its own preliminary motion (different per person, of course). Recording not only EMG during a static gesture pose but also the gestures including their preliminary motion may boost the credibility of the predictions.

However, in order to achieve this, the calibration/recording needs to sample the user data more interactively. However, if it is achieved, we can compare the with and without preliminary motion classifier model performances.

No file, test, or entry point is identified in the issue. Start by locating the existing calibration and recording flow, then define interactive sampling for preliminary motion and compare classifier performance with and without it; done means both data-collection modes and their comparison are evaluated.

Written by the indexing model from the issue text.