document that DenseArrays do not need to define strides to follow the strided array API

[LilithHafner]

The true default definition, as per @less strides([]) is

strides(a::Union{DenseArray,StridedReshapedArray,StridedReinterpretArray}) = size_to_strides(1, size(a)...)

But I wrote an equivalent default that does not depend on internals.

[jishnub]

I find the title a little confusing, since

julia> DenseArray <: StridedArray
true

Are StridedArrays not guaranteed to have strides defined for them?

[LilithHafner]

Are StridedArrays not guaranteed to have strides defined for them?

From reading the docs... no? The only thing I can see as guidance for declaring a new type <: DenseArray is

The elements of a dense array are stored contiguously in memory.

from the docstring of DenseArray. Maybe we should add a fallback definition for strides(a::StridedArray) or at least strides(a::DenseArray) based on eltype and size.