Performance of `didRenameFiles` for Rascal

[toinehartman]

Describe the bug

The current implementation for computing rename edits for moved modules is quite slow, since it iterates over all qualified name instances. We should revisit this.

Some ideas:

• A substring-based initial search, which discards most files, before parsing and visiting the tree (similar to what rename does).
• Apply edits in multiple stages; module headers first, qualified references after. Apply edits to files that are open first?
• Normalization of (redundantly) escaped qualified names should also be considered carefully.

Expected behavior
Fast and responsive refactorings.

[toinehartman]

Before doing this, we should probably first invest in some tests (#848).