SuperSuperSorter

[lpgaff]

The latest sorting upgrade, now using smart pointers inside the vector so that we don't need to move the data around when we sort it, only the pointers to the memory. Tested with 50 sub runs and it is significantly faster and doesn't appear to be leaking memory.

For a single subrun as comparison:

Super^2-sorter = 14.24s user 1.79s system 93% cpu 17.082 total
Super-sorter   = 36.81s user 2.53s system 97% cpu 40.183 total
Root sorter    = 94.80s user 3.54s system 99% cpu 1:38.86 total

A factor of a 6.6 times faster overall.

This PR also includes Nigel's catch for corrupted input files in Issue #58.

[lpgaff]

Super^2-sorter = 14.24s user 1.79s system 93% cpu 17.082 total
Super-sorter   = 36.81s user 2.53s system 97% cpu 40.183 total
Root sorter    = 94.80s user 3.54s system 99% cpu 1:38.86 total

Of course, there was a bug in the time ordering and it wasn't doing it correctly... Real result is:

Super-sorter^2 = 36.67s user 3.90s system 97% cpu 41.436 total
Super-sorter   = 36.81s user 2.53s system 97% cpu 40.183 total
Root sorter    = 94.80s user 3.54s system 99% cpu 1:38.86 total

So not much different than before.

[lpgaff]

With the bugs fixed and using a mapping of data_id in the vector with timestamp, rather than calling it from the DataPackets class each time, we win another few seconds:

Super^2-sorter = 29.17s user 8.82s system 97% cpu 39.152 total
Super-sorter   = 36.81s user 2.53s system 97% cpu 40.183 total
Root sorter    = 94.80s user 3.54s system 99% cpu 1:38.86 total