Code looks good!

BTW, why did you "Replace ExtendedMatrix and DataColumn types with MatrixEntry container."? It seems like this was not necessary. No strong opinion, just wondering. I kinda liked ExtendedMatrix as a type for the full matrix.

why did you "Replace ExtendedMatrix and DataColumn types with MatrixEntry container."?

Good question. I actually wanted to keep the types, but ran into remerkable issues. I tried doing this:

- List[Cell, MAX_CELLS_IN_EXTENDED_MATRIX]
+ List[Tuple[Cell, KZGProof], MAX_CELLS_IN_EXTENDED_MATRIX]

And it had this error:

Spent some time trying to fix it but couldn't figure it out.

why did you "Replace ExtendedMatrix and DataColumn types with MatrixEntry container."?

Good question. I actually wanted to keep the types, but ran into remerkable issues. I tried doing this:

- List[Cell, MAX_CELLS_IN_EXTENDED_MATRIX]
+ List[Tuple[Cell, KZGProof], MAX_CELLS_IN_EXTENDED_MATRIX]

And it had this error:

Spent some time trying to fix it but couldn't figure it out.

Hmm, might be worth looking into fixing this.

Right now DataColumn is featured prominently in p2p-interface.md as well. So if we stick with the "untyped" approach of this PR, we should think of how to revise the p2p layer.

• I don't have a strong feeling about making ColumnIndex an alias of CelIID by inheritance

I don't have a strong opinion on this either. Maybe we should wait a little while and/or do it in a different PR.

• +1 on renaming CellID to CellIndex

Great, agreed! Unless you want me to do it here, I would like to do this in another PR. It'll be easy, just a lot of changes.