Wrong calculation of convergence to state if distance to state is exactly 0.0

[MarvynBornemann]

In Observables.jl the function

function get_convergence_to_state(sol::AbstractODESolution, state, distance; tail_frac=0.8)
    L = length(sol.t)
    idx = max(1, round(Int, tail_frac*L)):L
    x = sol.t[idx]
    y = [distance(state, p) for p in sol.u[idx]]
    X = zeros(length(x),2)
    X[:,1] = x
    X[:,2] .= 1.0
    slope, intercept = X \ log.(y)
    return slope
end

returns slope = NaN if the distance(state, p) = 0.0 and therefore y is 0.0 somewhere.

Might be resolved by just calculating slope = x \ log.(y). Then the slope = -inf if the distance y is 0.0 somewhere.