Fix #53 with ismutable default

[Tokazama]

This makes it so that ismutable has a default now, which is to use the mutable field of a type.

[ChrisRackauckas]

I'm not sure we should do this. It's almost guaranteed to not be correct for wrapper types.

[Tokazama]

It doesn't change any of the current behavior for wrapper types. It simply provides a fall back. If we want to account for wrappers I could replace the current ismutable(::Type{<:AbstractArray}) = true with

function ismutable(::Type{T}) where {T}
    if parent_type(T) <: T
        return T.mutable
    else
        return ismutable(parent_type(T))
    end
end

...and then we could explicitly define the mutability of each type.

[Tokazama]

I added the recursive version that searches for a parent structure. I still haven't changed any of the tests, because the behavior is essentially the same.

[Tokazama]

@ChrisRackauckas, is this PR still problematic?

[ChrisRackauckas]

I think that works.

[ChrisRackauckas]

Could it get a test? Fake wrapper

[Tokazama]

So instead of testing a fake wrapped array I just tested views of StaticArrays. Turns out this actually fixes views on immutable arrays (which previously just assumed mutability).

[ChrisRackauckas]

I'm convinced.