Description

Our current scanner implementation scans the entire project, including node_modules (if project rules are enabled) and files that are ignored using files.includes (but not entire directories). While this is mostly intentional, it is far from optimal.

Instead, we should scan all the files listed in files.includes and then (if project rules are enabled) descend into dependencies, but only scan the dependencies that are reachable from files.includes.

Eventually the goal should be that we can deprecate files.experimentalScannerIgnores.