The summary parsing code now seems to handle paired end runs

The summary parsing code now seems to handle paired end runs
this required changing how the lane_results were being stored,
previously it was a dictionary indexed by lane, now it is a list
of dictionaries, where the list index indicates which "end" of
a paired end run it is. (0 is the first, 1 is the second)

Also I got tired of being forced to use strings for the lane index
by element tree and modified the code so it converts the strings
required by element tree to integers for our internal dictionaries.