Ok so for now this works, although I am surprised there isn’t a way to store idiosyncratic Julia structs:
using CSV, Tables
to_store= reduce(vcat, A) # as big matrix
CSV.write("FileName.csv", Tables.table(to_store), writeheader=false) #easily storable in universal formats
temp= CSV.File("FileName.csv", df) |> Tables.matrix() #call and transform to matrix
#shape into arrays of arrays.
to_use= [temp[i:(i+size(temp,2)-1), :] for i in range(1, size(temp,1), step= size(temp,2)]
> to_use == A
true