Drop "ifdef GHC"s on INLINABLE pragmas

[meooow25]

These were added long ago in 06e91ac as an improvement on the already existing INLINE pragmas. We can get rid of these unnecessary (and confusing to contributors) ifdefs today since we do not have other compilers where we want the non-INLINABLE behavior to apply.

[meooow25]

Motivated by #1163 (comment)

[treeowl]

Aren't they there because INLINABLE isn't a Report standard feature?

[meooow25]

That's a good point to consider, here's the report on pragmas: https://www.haskell.org/onlinereport/haskell2010/haskellch12.html

An implementation is not required to respect any pragma, although pragmas that are not recognised by the implementation should be ignored.

So, if the non-GHC compiler follows the report there should be no problems with unguarded INLINABLE pragmas.

[treeowl]

They'll very likely throw up a slew of warnings 🫤. That hypothetical problem is probably worth incurring to remove the actual clutter though.

[meooow25]

Yes, I also think cleaning up the clutter would be nice. For what it's worth, MHS is the only such compiler at the moment and it's not complaining.