The main goal of this patch is that the column widths and ordering is not reset
every time one switches between different contexts.
It does that by keeping track of multiple `SpreadsheetTable`. There is one for
each table that is viewed (so e.g. the point and edge domain of the same mesh
are two different tables). Each table has an identifier and an array of columns.
There is some garbage collection in place so that the number of stored tables
does not increase unbounded.
This also comes with an updated Python API:
```python
import bpy
spreadsheet = bpy.context.screen.areas[...].spaces.active
active_table = spreadsheet.tables.active
print(active_table.id.type)
print(active_table.id.attribute_domain)
print(active_table.columns[0].id.name)
```
In the future, we might add some smarter logic to keep tables with different
identifiers more in sync. We don't have a great heuristic for that yet.
Pull Request: https://projects.blender.org/blender/blender/pulls/139205
0b24f15939