Convergence timeseries for lyapunov functions

[Datseris]

After Julia 0.7 with IPO we will be able to return convergence timeseries instead of final values without having to re-write code, while still ensuring type stability only by function arguments.

At the moment it is trivial to get the lyapunov convergence timeseries, by duplicating the main function lyapunovs and change this: λ += log.(abs.(diag(R)))
to this: push!(λ, λ[end] + log.(abs.(diag(R)))
and add 2 more lines:
λ = T[]; tvec = T[];
and inside the loop put push!(tvec, t).

It is literally 3 lines of code but it is very not elegant. With the new version it may be possible to do this changes without duplicating code and while still keeping things transparent and concise.

[Datseris]

I am also wondering on what is the take on always returning convergence timeseries, instead of final value.

What is the consencous on that? I have not done it like this because it is a tiiiiiiiiny bit faster to not create vectors, in case the user doesn't want them. But apparently most users actually want the lyapunov convergence timeseries instead of just a final value.

So maybe it is not worth it to support both options, and just go on and always return convergence timeseries instead.