Should `printchplenv` print `LLVM_SUPPORT` by default? (and maybe other `LLVM` settings as well?)

[bradcray]

On a few occasions, users have expressed confusion that we're trying to make use of LLVM when they have CHPL_LLVM=none set. As developers, we know that the reason is that we're now using the LLVM support libraries for the compiler build, but this isn't generally well-known, and it's certainly understandable why our trying to use LLVM when CHPL_LLVM=none could be counterintuitive.

A simple step we could take to try and address this would be to print the LLVM_SUPPORT setting by default with printchplenv as well. This would at least give some indication that even with CHPL_LLVM=none, there are other settings that involve LLVM and are defaulting to something else.

(For that matter, maybe we should consider printing all of the LLVM variables by default given that determining what LLVM installation someone's using has become one of the most common things for us to debug when users have build questions?)

(We could also consider renaming CHPL_LLVM to something that's more specific about what exactly it does—like CHPL_LLVM_BACKEND, though this isn't quite accurate since it's also involved in the front-end for parsing extern blocks… But that's a bigger change and involves design discussions. The benefit would be that it arguably wouldn't be as surprising that a CHPL_LLVM=none build would try to use LLVM).