Performance Issues with layer_likelihood (sparse matrix representation)

[IngoScholtes]

From pathpy created by verginer : sg-dev/pathpy#50

The layer_likelihood function in tests has a bottleneck the indexing of the transition matrix
https://github.com/sg-dev/pathpy/blob/595fc2d497b7446112ad34c562582d57079f6656/pathpy/classes/multi_order_model.py#L473-L480

Some stats changing the type of sparse matrix used:

	dense	lil	csr	csc
layer_likelihood	2.8	3.5	10	27
estimate_order	16.8	17.6	24	42
with instantiation	29.1	30	41	51