reporting un-matched events

[nsteinme]

Events of high enough magnitude that didn't match any template should be reported. As an easiest thing, it could just tell you how many of these there were (on each channel, maybe), so you know if it's a problem. Second stage, it could report the event times and peak channels of all of these, perhaps even adding them to the clustering with an extra template for each channel or local group of channels, and give it a flat template shape. This way these events could be easily visualized in phy.

It might be tricky though - one might expect lots of residuals (events left after subtracting some spike) to be unmatched, but these would look worrying to the user since they will see the raw data, without the first spike subtracted.