Overhaul build system to reduce memory usage and speed up builds

[lifflander]

What Needs to be Done?

With cmake 3.16 released, we have a whole host of new methods for reducing compile times. This will include:

• Switching to the gold linker instead of ld for CI builds (at least) since it is known to be faster and less resource intensive
• Explore/implement pre-compiled headers now supported CMake
• Explore/implement unity builds to reduce compile times and memory usage

[lifflander]

The baseline time with the gold linker, full build, on my Mac with gcc-7 (code coverage enabled) with 8 cores dedicated to the docker container is. Size is 25 GiB.

baseline, gold, full build:
real    19m41.686s
user    117m0.984s
sys     22m27.988s

[lifflander]

With a Unity build enabled on just VT (libvt):

real    15m42.910s
user    67m12.630s
sys     9m53.687s

[lifflander]

Working on unity build for tests, which I think will have a huge benefit, but it's a pain in the ass.

[lifflander]

With the full unity build (and lots of changes to go along with them), down to:

real    13m44.297s
user    54m17.289s
sys     8m39.568s

And 18 GiB vs. 25 GiB