benchmark and performance optimization for selection range

[kokobd]

Is your feature request related to a problem? Please describe.

According to discussions in #2565, the current implementation doesn't have an optimal complexity. It is O(n) now, but could be done in O(log(n)). We may add a benchmark to check what will happen when requesting selection ranges in a very large file.

Describe the solution you'd like

• Add a benchmark (currently there is no benchmark for plugins, so this might become the first one).
• Implement a better algorithm (see comments in findSelectionRangesByPositions)

Describe alternatives you've considered

Maybe we don't need to optimize, as it works very fast for files with 1000 lines, and large files are rare in practice (why not split them into smaller modules?).

Additional context

The orignal feature request is #212

[kokobd]

This is partially solved by #3003, where I applied binary search on each level, and tree search on the whole AST (instead of the previous naive linear search). Theoretically, the performance should improve for large files.

I think it's ok to close this issue and open another one to add benchmarks if I have the energy to do that.