Add method to horizontally concatenate two (or more) tables of possibly different type

This has come up in ML workflows; see https://github.com/alan-turing-institute/MLJ.jl/issues/915. Would TableOperations.jl be the appropriate place for this?

What I have in mind is a simple concatenation - not a fancy join. So, if a column name of table1 appears in table2, then the table2 column just gets added with its name modified.

The tricky part is deciding on what the return type should be. I don't have fixed ideas about this, but perhaps if the tables do have the same type, and that is a sink type, then that is also the return type.

Although it is not a part of the public API, I see that TableTransforms.jl has an implementation. (To get the final table, the type of the first table is materialized.):

julia> table1
3×2 DataFrame
 Row │ x        z        
     │ Char     Float64  
─────┼───────────────────
   1 │ 𘂯       0.673471
   2 │ \U3f846  0.360792
   3 │ \Ud50cb  0.68075

julia> table2
(x = [0.41754294943943493, 0.7713462387833814, 0.9189998773436003], y = ['\U84fa1', '\U5e144', '\U872a4'])

julia> TableTransforms.tablehcat([table1, table2])
3×4 DataFrame
 Row │ x        z         x_        y       
     │ Char     Float64   Float64   Char    
─────┼──────────────────────────────────────
   1 │ 𘂯       0.673471  0.417543  \U84fa1
   2 │ \U3f846  0.360792  0.771346  \U5e144
   3 │ \Ud50cb  0.68075   0.919     \U872a4

cc @ExpandingMan @juliohm

Start by reading the linked MLJ.jl issue and the TableTransforms.jl implementation in src/transforms/parallel.jl, especially tablehcat. Then inspect the TableOperations.jl and Tables.jl interfaces to determine how heterogeneous inputs and duplicate column names should be handled. Done means a documented horizontal concatenation operation with an agreed return-type rule and coverage for the shown table forms.