add benchmarking of input/parsing and output phases

[gkthiruvathukal]

In the C version, I have added some code to break out the time by phases:

I/O time 300.913000
qsort time 26.050000
tree building time 60.848000
printing time 591.108000
Total elapsed time 979.010000

I will be making a couple of changes to how I label these, but I am leaning toward this:

I/O + parsing time 300.913000
qsort time 26.050000 <-- This only applies to the C version.
tree building time 60.848000 <-- you'd probably use the word map here.
process tree generation/printing time 591.108000
Total elapsed time 979.010000

I'm assuming all of this can be done with Scala meter. I think this would also help us to see how well various languages do with their idiomatic I/O. I've always felt that Java I/O gets a bad rap. With its buffered abstractions, it should do in theory. Ditto for Scala.