Improve performance of Problem.shouldSkip() for large, new projects

[RBusarow]

The current implementation re-parses the build file for every single finding. This is extremely time-consuming on huge projects with lots of findings.

Some observations:

• "inherited" findings can't really be suppressed as it is, since the declaration is for a different module (and there can be many). Either that suppression method should be updated, or the check should be skipped for that and any similar type.
• those annotations/comments won't be added mid-task, so the amount of parsing being done is redundant.
  • One fix could be to scan each build file once for the annotations/comments, and set a flag for that file once to say if it can be skipped.