Experimental support for loading JIMAGE (JDK 9+) files

[RyanGlScott]

This bumps the crucible submodule to include GaloisInc/crucible#638, which adds basic support for handling JDK 9 or later. JDK 9+ packages its standard library not in a JAR file, but in a JIMAGE file. For more details on how crucible-jvm handles JIMAGE files, refer to Note [Loading classes from JIMAGE files] in Lang.JVM.Codebase.

This fixes #861, although there are still unsolved issues that arise when using modern JDKs with certain classes, such as String. As a result, I have decided to label support for JDK 9+ as experimental:

• I have updated the SAW documentation to mention these shortcomings.
• I have opened crucible-jvm: Investigate bugs when verifying JDK 9+ code that uses String crucible#641 to track the remaining issues.

Other things:

• crucible-jvm: Add --java-bin-dirs crucible#636 and Lang.JVM.Codebase: experimental support for loading JIMAGE files crucible#638 upstreamed the code from SAWScript.JavaTools and SAWScript.ProcessUtils into crucible-jvm, so we can remove these modules in favor of importing Lang.JVM.JavaTools and Lang.JVM.ProcessUtils from crucible-jvm.

• I removed the dependency on the xdg-basedir, as it was unused. This dependency was likely added quite some time ago, and it appears that saw-script switched over to using XDG-related functionality from the directory library since then. I opted to use directory to find the .cache directory as well, so I have made that clear in the .cabal file.

• The biJavaCodebase :: Codebase field of BuiltinContext is completely unused, which I noticed when making changes to the Codebase type. Let's just remove it.

[RyanGlScott]

I've rebased this on top of #1005, so the diff is slightly more amenable to reviewing than it was before. I'm still relying on a hack (copy-pasting the entire definition of Codebase and related functions in SAWScript.JavaCodebase) to avoid having to change any submodules, but you may find it easier to review what the changes look like if they were upstreamed in crucible-jvm.

[RyanGlScott]

I've rebased this on top of GaloisInc/crucible@e1531a0, and I have updated the documentation to make it clear that support for JDK 9+ is experimental. This is now ready for review.

I briefly considered if saw should emit a warning every time a user tries running saw with JDK 9+ or later (much like GaloisInc/crucible@e1531a0 does), but in the end, I decided against it. It's possible that someone could be running saw solely to verify LLVM code, but if they have JDK 9+ on their PATH, then they'd get annoying warnings about a SAW backend that they're not using. We could envision a more complex way to track if Java code is being verified—for example, we could have an IORef Bool that tracks if a JVM-related SAW function has been invoked before, and if such a function is invoked, emit a warning and change the IORef Bool such that we don't emit the warning more than once per SAW invocation. The amount of elbow grease required to get that working didn't seem to justify the payoff, so I simply updated the documentation and called it a day.