kill 15 more PLT entries

[Roland McGrath]

> I believe malloc.c itself, if you are not compiling with -O3.

Ah.  Well, since this PLT reduction is nothing but hyperoptimization anyway
I am inclined to think that it's not worth extra hoops to make it a little
better when you are not optimizing as much as can be done.  Every one of
these little things further obfuscates the code, so I think there is a cost
to it that should be outweighed by material performance benefit.  If there
is no benefit except when you are not building for the best performance
readily available, not worh it.