Should subtypes of `Factorization` maintain keys?

[rofinn]

I'm wondering if we should try to maintain the keys on the components of Factorization types like SVD and Cholesky? I realize it wouldn't apply to all components, but for upper and lower matrices it might be a reasonable to avoid prematurely dropping keys?

[mcabbott]

It could do, I'm not opposed but haven't run into a use for this myself.

In https://github.com/invenia/NamedDims.jl/pull/24 this was going to be done by overloading getproperty: https://github.com/invenia/NamedDims.jl/pull/24/files#diff-11fb1118a11e25a31717086efe376fa8R101-R113 But it would also be possible to construct an SVD object in which the wrappers are already attached, I think.