`HaarAverageFilter` breaks memory contiguity when broadcasted

[Yuan-Ru-Lin]

Here's a MWE

using RadiationDetectorSignals, Unitful, ArraysOfArrays, LegendDSP

wvfs = ArrayOfRDWaveforms((fill(range(0f0, 990f0, 100) .* u"ns", 10000), nestedview(rand(100, 10000))))
wvfs2 = wvfs .|> HaarAveragingFilter(2)

println(typeof(wvfs.signal))    # ArrayOfSimilarArrays{Float64, 1, 1, 2, Matrix{Float64}}
println(typeof(wvfs2.signal))   # Vector{Vector{Float64}}

[oschulz]

Wanna try your hand at a PR to fix it?

[Yuan-Ru-Lin]

I am working on a supervised learning study with Geant4.jl but I believe I will have to fix this if I proceed far enough; that is, I might do DWT at some point. But for now I need to make sure there's a way to identify SSEs and MSEs based on MC information