Incorrect handling of `view` kwarg in Tables.subset

[bkamins]

Example:

julia> nt = (a=[1,2,3], b=[4,5,6])
(a = [1, 2, 3], b = [4, 5, 6])

julia> x2 = Tables.subset(nt, 1, view=true)
(a = 1, b = 4)

julia> x1 = Tables.subset(nt, 1, view=false)
(a = 1, b = 4)

julia> nt.a[1]=100
100

julia> x2
(a = 1, b = 4)

julia> x1
(a = 1, b = 4)

vs

julia> df = DataFrame(a=[1,2,3], b=[4,5,6])
3×2 DataFrame
 Row │ a      b
     │ Int64  Int64
─────┼──────────────
   1 │     1      4
   2 │     2      5
   3 │     3      6

julia> x1 = Tables.subset(df, 1, view=false)
Tables.ColumnsRow{DataFrames.DataFrameColumns{DataFrame}}: (a = 1, b = 4)

julia> x2 = Tables.subset(df, 1, view=true)
Tables.ColumnsRow{DataFrames.DataFrameColumns{DataFrame}}: (a = 1, b = 4)

julia> df.a[1] = 100
100

julia> x1
Tables.ColumnsRow{DataFrames.DataFrameColumns{DataFrame}}: (a = 100, b = 4)

julia> x2
Tables.ColumnsRow{DataFrames.DataFrameColumns{DataFrame}}: (a = 100, b = 4)

So we see that view kwarg when returning row is not respected and is not handled consistently.

[bkamins]

See JuliaData/DataFrames.jl#3158 for what I believe is a concrete implementation of Tables.subset (an example for data frame)