Support missing values

[haberdashPI]

This does not seem to work:

PaddedView(missing,rand(10,2))

Would be useful for computing various statistics, over an iterator of arrays, where I want to skip missing values if they fall outside the range of an individual array.

[johnnychen94]

This is actually supported, note that rand(10, 2) returns Array{Float64, 2}, which doesn't accept missing values.

julia> x = Any[1 2 3;4 5 6]
2×3 Array{Any,2}:
 1  2  3
 4  5  6

julia> PaddedView(missing, x, (-3:3, -3:3))
7×7 PaddedView(missing, ::Array{Any,2}, (-3:3, -3:3)) with eltype Any with indices -3:3×-3:3:
 missing  missing  missing  missing   missing   missing   missing
 missing  missing  missing  missing   missing   missing   missing
 missing  missing  missing  missing   missing   missing   missing
 missing  missing  missing  missing   missing   missing   missing
 missing  missing  missing  missing  1         2         3
 missing  missing  missing  missing  4         5         6
 missing  missing  missing  missing   missing   missing   missing

[haberdashPI]

Right, but padded view could be smart enough to change the eltype to Union{Missing,Float64} even if the underlying array has an eltype of Float64.

[timholy]

Want to try a PR, @haberdashPI?