Explore possibilities to combine leadfield dimensions for computing the magnitude

[jschepers]

At the moment, the magnitude can be computed given the orientation of a headmodel or a custom orientations vector. It would also be nice to provide more possibilities of how the leadfield dimensions could be combined e.g. L2 norm, sum, maximum etc.

@behinger visualized two possibilities and also suggested: SVD/largest eigenvalue, PCA, monte-carlo, norm, perpendicular.

Here is the code for computing the norm which was part of UnfoldSim before:

function magnitude(lf::AbstractArray{T,3}) where {T<:Real}
    si = size(lf)
    magnitude = fill(NaN, si[1:2])
    for e = 1:si[1]
        for s = 1:si[2]
            magnitude[e, s] = norm(lf[e, s, :])
        end
    end
    return magnitude
end

[behinger]

I just pushed a PR #131 with a explanation/exploration discussing this.

Btw. the code can be greatly simplified:

m_norm = mapslices(norm, L; dims = (3))

haven't checked the tutorial if it actually runs - because I'm running out of time 🙈