Fix empty StridedView broadcasting

[mtfishman]

This fixes a bug I came across when broadcasting empty StridedViews. Without this change, I see the following:

(dev) pkg> st
Status `~/.julia/dev/Strided/dev/Project.toml`
  [5e0ebb24] Strided v2.3.4

julia> using Strided

julia> A1 = StridedView(randn(2, 0))
2×0 StridedView{Float64, 2, Memory{Float64}, typeof(identity)}

julia> A2 = StridedView(randn(2, 0))
2×0 StridedView{Float64, 2, Memory{Float64}, typeof(identity)}

julia> A1 .+ A2
ERROR: ArgumentError: step cannot be zero
Stacktrace:
  [1] steprange_last
    @ ./range.jl:349 [inlined]
  [2] StepRange
    @ ./range.jl:336 [inlined]
  [3] StepRange
    @ ./range.jl:391 [inlined]
  [4] _colon
    @ ./range.jl:24 [inlined]
  [5] Colon
    @ ./range.jl:22 [inlined]
  [6] macro expansion
    @ ~/.julia/packages/Strided/yfmeB/src/mapreduce.jl:421 [inlined]
  [7] _mapreduce_kernel!(f::Strided.CaptureArgs{…}, op::Nothing, initop::Nothing, dims::Tuple{…}, blocks::Tuple{…}, arrays::Tuple{…}, strides::Tuple{…}, offsets::Tuple{…})
    @ Strided ~/.julia/packages/Strided/yfmeB/src/mapreduce.jl:257
  [8] _mapreduce_block!(f::Strided.CaptureArgs{…}, op::Nothing, initop::Nothing, dims::Tuple{…}, strides::Tuple{…}, offsets::Tuple{…}, costs::Tuple{…}, arrays::Tuple{…})
    @ Strided ~/.julia/packages/Strided/yfmeB/src/mapreduce.jl:170
  [9] _mapreduce_order!(f::Strided.CaptureArgs{…}, op::Nothing, initop::Nothing, dims::Tuple{…}, strides::Tuple{…}, arrays::Tuple{…})
    @ Strided ~/.julia/packages/Strided/yfmeB/src/mapreduce.jl:154
 [10] _mapreduce_fuse!
    @ ~/.julia/packages/Strided/yfmeB/src/mapreduce.jl:130 [inlined]
 [11] copyto!
    @ ~/.julia/packages/Strided/yfmeB/src/broadcast.jl:39 [inlined]
 [12] copy
    @ ./broadcast.jl:919 [inlined]
 [13] materialize(bc::Base.Broadcast.Broadcasted{Strided.StridedArrayStyle{…}, Nothing, typeof(+), Tuple{…}})
    @ Base.Broadcast ./broadcast.jl:894
 [14] top-level scope
    @ REPL[9]:1
 [15] top-level scope
    @ REPL:1
Some type information was truncated. Use `show(err)` to see complete types.

Other codepaths like map[!] seem to already have this kind of check for the empty case (before calling _mapreduce_fuse!).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/broadcast.jl	89.28% <100.00%> (+89.28%)	⬆️

... and 6 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[lkdvos]

Thanks for the report and fix! (And added tests, greatly appreciated)