I usually edit and build many times before I'm happy enough with my work to even try "make test" (or the Windows equivalent :-). Which of these options *compiles* the fastest? Maybe that one should be the default. But perhaps other peoples' workflow is different (in fact it would be surprising if we all had the same workflow for writing C code :-).

In fact I'm guilty of not even building using --with-pydebug until I run into a problem I can't debug with fprintf().  :-) But once I do, it's likely that I want to aim a debugger at the code, *and* I'd be recompiling repeatedly. The time to complete a full "make test" run on my local machine doesn't bother me, it's too slow to have in my development loop either way.