Reinstate the time histogram of events

An early iteration of the Tristan tools included a time histogram tool, which disregarded the `event_id` information and simply binned events into time bins, equivalent to what you would get by summing the counts on each image from `images multi` or `images pp`. It is obviously a trivial calculation to `bincount` the image numbers, so this should be reinstated.

Output was a PNG plot of the histogram and a CSV table of bin edges and frequencies, simply created with Pandas. This seemed to provide useful functionality to I19 beamline staff and users at the time.