More glib benchmarking

[Ulric Eriksson <ulric edu stockholm se>]

I have made another memory management benchmark, this time to test what
happens when glib is made to handle more than one data type. Or rather
verify, because I was pretty sure what should happen: glib would be
slightly faster and use less memory for one datatype, due to avoiding
some malloc overhead. When more than one type was involved, glib would
still be faster but less efficient in terms of size.

I was right on one out of three systems.

Details are available for those who are interested at

	http://www.edu.stockholm.se/~ulric/glib/mem-comparison.html

Not here, to protect those who really couldn't care less.

Ulric