Converted assembly to use gcc instead of nasm, updated html man

Converted assembly to use gcc instead of nasm, updated html man
pages.
The x86 assembly loops is now a single option to configure,
and the single/double prec. is controlled with --enable-float
(default is yes), to be consistent with fftw.
Removed the less common options from the summary printed by
configure, but they are still available.
Introduced libtool to create both static and dynamic libraries -
you can control it with configure options. --disable-shared might
be suitable for development work.
To avoid compiling both PIC and non-PIC code you can try --with-pic,
but the default is both.