An initial work on a dependency tree for formulas

[jahav]

The tree can mark formulas as dirty, if some precedent (direct/transitive) cell is changed.

The chosen data structure is an R-tree. R-tree seems to be the only reasonable structure (other than the variant R*-tree) for orthogonal range searching for overlapping ranges, most other are point vs range or "variants" of the main algorithm that is intended for other things (e.g. kd-tree).

R-tree has an implementation in form of RBush library. It isn't strongly named and won't be, will need to make a fork or abandon strong naming.

This is only initial work, will need to be improved to deal with

• adding/removing formulas in sheets
• name dependencies
• include to workbook (currently is apart and unused)
• workbook modifications (adding/renaming/removing sheet, adding rows, removing rows, shifting) - will likely just recreate from scratch if needed when it happens. There is no time do it more accurately currently.