deleteat! where drop is a column

[gustafsson]

deleteat!(df, df.what_to_drop) is broken unless df.what_to_drop is the last column.

julia> df = DataFrame(a=[false, true, true], b=1:3)
3×2 DataFrame
 Row │ a      b     
     │ Bool   Int64 
─────┼──────────────
   1 │ false      1
   2 │  true      2
   3 │  true      3

julia> deleteat!(df, df.a)
ERROR: BoundsError: attempt to access 3-element Vector{Int64} at index [false]
Stacktrace:
 [1] deleteat!(a::Vector{Int64}, inds::Vector{Bool})
   @ Base ./array.jl:1626
 [2] _deleteat!_helper(df::DataFrame, drop::Vector{Bool})
   @ DataFrames ~/.julia/packages/DataFrames/LteEl/src/dataframe/dataframe.jl:922
 [3] deleteat!(df::DataFrame, inds::Vector{Bool})
   @ DataFrames ~/.julia/packages/DataFrames/LteEl/src/dataframe/dataframe.jl:894
 [4] top-level scope
   @ REPL[40]:1

This PR creates a copy of which rows to drop when needed to avoid this.

[bkamins]

@nalimilan - OK to merge?

[nalimilan]

This seems like an extremely unlikely situation, but better be correct in that case of course. :-)

[bkamins]

Thank you!