Switch renamecol to rename and insertcol to insertcols, add multiple column version

[piever]

This is the last part of #239, renaming renamecol to renamecols and deprecating insertcol to insertcols plus multiple columns version.

Two doubts:

1. Should the name be renamecols or just rename? After all we are obviously renaming columns (it wouldn't make to rename rows). Maybe we should just be consistent with DataFrames and call it rename.

2. I've interpreted the "multiple argument" version of insertcols! to behave as follows:

julia> t = table(rand(3), rand(3), rand(3), names = [:a,:b,:c])
Table with 3 rows, 3 columns:
a         b         c
────────────────────────────
0.509087  0.763431  0.557554
0.217274  0.801138  0.160908
0.462436  0.506603  0.921267

julia> insertcols(t, 2, :d => rand(3), :e => rand(3))
Table with 3 rows, 5 columns:
a         d         e         b         c
────────────────────────────────────────────────
0.509087  0.212275  0.683083  0.763431  0.557554
0.217274  0.49579   0.516921  0.801138  0.160908
0.462436  0.404024  0.982535  0.506603  0.921267

julia> insertcolsbefore(t, :c, :d => rand(3), :e => rand(3))
Table with 3 rows, 5 columns:
a         b         d          e         c
─────────────────────────────────────────────────
0.509087  0.763431  0.730766   0.263156  0.557554
0.217274  0.801138  0.844378   0.915973  0.160908
0.462436  0.506603  0.0846812  0.163381  0.921267

That is, we pass a unique index and all columns are inserted starting from there. I thought the version with multiple indices would be too confusing as the indices would change position when the first columns are inserted. Does this make sense?

[joshday]

1. I vote for rename.
2. Does it make sense? Yes.

[piever]

OK, I'm going with rename (simpler and in agreement with DataFrames).