Fix grouping and reduce RAM usage

[shubhra-agrawal]

Copied from an email received from @AndrewRosko: I have indeed found a small bug (2 missing lines) in the EIC pulling code that causes incorrect grouping of peaks, especially when searching by retention time as well as m/z.

The only other change I have made so far is another 2-line change in the mzSample file loader that trims some zeros off the Scan arrays, to cut the RAM usage by ~50%. On our system, despite still loading the entirety of the intensity data into memory, this makes a big difference in file load time.