REPL: JLine: follow recommendation to use JNI, not JNA; also JLine 3.27.1 (was 3.27.0)

[SethTisue]

as per the https://github.com/jline/jline3 readme

and as per discussion and linked items on #22201

fixes #22201

note that as far as I can tell, the stuff I removed from libexec/common-shared is dead code

@philwalk dunno if you're still around but judging from #12405 you might be a good reviewer here

note that I believe we don't need to also port scala/scala#10889 here, since we are already using separate JLine JARs rather than the all-in-one JAR

I've chosen not to upgrade all the way to JLine 3.28.0 at the moment, as it is quite new (2 days ago) and doesn't appear to have any fixes that might be critical.

[SethTisue]

I am not a good person to test this on Windows, as I don't currently even have a working Windows setup... and even if I did, I know nothing about cygwin/conemu/mingw/???, it's just meaningless noises to me

I've added the "backport:nominated" label, but it probably shouldn't be backported to LTS too soon? Might be better to ship it in a 3.6.x release first and make sure the Windows users don't take up torches and pitchforks in response?

We are more or less YOLOing this in both sbt and Scala 2.13.16, so we might get problem reports from either of those, too. cc @Friendseeker

Regardless, I don't think we need to be too frightened, as JLine is in wide use outside of Scala as well.