cdecl99: Optimize print_version a bit.

Instead of a whole bunch of separate print calls, convert print_version
to a single call to a printing function, with a single string argument.
Compilers do not seem to coalesce multiple consecutive prints together,
so doing it explicitly gives a modest reduction in code size.

Wrap that all up into a new set of macros so that the same trick can
also be shared with the test applications.