compiling CFEMM on Mac M1

[ericmaslen]

When I compile CFEMM on my Mac M1 using
cmake .
make
I get no compiler errors after fixing sincos -> __sincos and adding malloc.h to the necessary
directory.

However, when I run
fmesher TorqueBenchmark.fem
it just goes into an infinite loop.

After a lot of debugging, I've determined that mergehulls is getting passed a defective innerleft otri argument by divconqrecurse: the pointer itself is non-NULL and it has a valid orientation field but the tri fields appear to be NULL pointers. The result is that the walk gap routine never converges (since the vertices are all NaN).

I'm a bit at wit's end: is this a problem with the assumptions about bit boundaries that are baked into the otri structure? I definitely don't want to drag myself into re-writing triangle.c!

Could I side-step this by using the newer substitute for triangle?