Fix switch/flip error rate calculations

[JosephLalli]

The previous code counted a string of errors "0 1 1 1 0" as two flips, two switch errors, and one correct variant. For researchers who want to examine the relative rate of flip events and switch errors unrelated to flips, this is unhelpful.

Instead, I would argue this represents one flip event, three switch errors, two correct variants, and zero "pure switch errors" (ie isolated switch errors that are not associated with a flip error).

This code change calculates flip events, switch errors, pure switch errors, and correct variants. It also calculated how many flip errors occur consecutively (ie, more than two errors in sequence).

Do you agree that this system of tracking errors and flips makes sense? I plan on publishing soon with data calculated using this code, and your sign off on this fork would be of immense help.

Thanks,
Joe