Handle Map.Entry properly in import removal, and remove duplicated logic

[kelloggm]

Prior to this PR, there were two pieces of code in Specimin that were both trying to handle removing extraneous imports:

• some code in PrunerVisitor, and
• the UnusedImportRemoverVisitor class

I deleted the logic in PrunerVisitor, which was more complex but had the same overall impact. I enhanced the logic in UnusedImportRemoverVisitor to fix the bug evinced by the new test case, wherein the import java.util.Map was incorrectly removed. I also had to add some continue statements elsewhere in PrunerVisitor; that part makes me nervous, because a seemingly-unrelated test failed without them but did not fail until I removed the logic in PrunerVisitor. However, I much prefer this design and don't have time to dig deep into that, so I'm okay with ignoring that weirdness for now: the new continue statements are pretty clearly correct (it's not sensible to keep going through the loop body's logic if we've already decided to remove something).

I expect CI to fail, because this PR should resolve some compilation issues in plume-util. I'll update the "passed" baseline number once I know what the new result is.

[M3CHR0M4NC3R]

I am able to reproduce these results on my local machine as well. The accuracy was also reported at 91.24. The accuracy file will conflict with the other pull requrests so that will need to be checked again afterwards. Otherwise this looks good to me.

[kelloggm]

I am able to reproduce these results on my local machine as well. The accuracy was also reported at 91.24

Thanks for confirming. This is good to go, in that case.