In the context of PR 234495 USE_GCC was described in the Porters Handbook,
including a note that it's a build *and* run-time dependency.
*) Since then I have implemented USE_GCC=yes:build, USE_GCC=10+:build, and
the like, which are worth mentioning.
*) And recommending USE_GCC=any as "typical" really is the opposite of what
I, and Mk/bsd.gcc.mk, have been recommending for years and years!
USE_GCC=any means that the port hardly ever is tested with base GCC, since
on amd64 it is equivalent to USE_GCC=yes and building with base GCC (if
present) is only ever tested on architectures that only very few people have.
Can you please change this to presenting USE_GCC=yes as the default and
noting the ":build" option?
I put effort into keeping Mk/bsd.gcc.mk up-to-date, and would be happy
for you to lift the description there and bring things more in sync again.