Implement recursive columnaccess and rowaccess

StockAndFlowF is currently stored as a NamedTuple of ACSetTables. NamedTuples such that T <: NTuple{N, AbstractVector} implement columnaccess, and ACSetTable implements column access, but this doesn't work recursively; that is, when accessing the tables of a StockAndFlowF, one columnaccess is false. If I understand correctly, though, it wouldn't be wrong to define it to be true for NamedTuples of ACSetTables or ACSetTables of NamedTuples; it simply means that accessing it is O(1), and we can just define columns(x) = x.

Intention is to be able to manipulate StockAndFlowF tables using [Table Operations](https://github.com/JuliaData/TableOperations.jl).

It may fall on Catlab.DenseACSets to implement this instead.

I've implemented a bit of a workaround which works specifically with NamedTuples containing ACSetTables:

const NamedACSetTable = NamedTuple{names, T} where {names, T <: NTuple{N, ACSetTable} where N}
Tables.columnaccess(x::NamedACSetTable) = true
Tables.columns(x::NamedACSetTable) = x

Tables.rowaccess(x::NamedACSetTable) = true