Incorrect interpretation of time shaped RFs

[beorostica]

In the source code we use a custom trapezoidal function (used only by the get_rfs() function) which doesn't regard time-shaped waveforms:

⏢(A, t, ΔT, ζ1, ζ2, delay) = begin
    if sum(abs.(A)) != 0 && ΔT+ζ1+ζ2 != 0 # If no event just ignore calculations
        #Getting amplitudes, only supports uniformly sampled waveforms for now
        if length(A) != 1
            grad_raster = ΔT / length(A)
            idx = ceil.(Int, (t .- delay .- ζ1) ./ grad_raster ) #Time to integer index
            valid = 1 .<= idx .<= length(A)
            idx[(!).(valid)] .= 1
            B = A[idx] .* valid
...

In a near future, we want to use linear interpolation instead of zero-order-hold so this issue should be addressed there. This is also related with the issue #167.

[cncastillo]

This already in development in the branch:

• https://github.com/cncastillo/KomaMRI.jl/tree/block-time-interpretation