Add per-event edge granularity besides per-edge granularity to CIF data-based synthesis

From #520 (comment 1114569):

[...]

Transition relation granularity

So far the workset algorithm and its parameters. Then one important thing remains. The transition relation is of a granularity that there is exactly one transition per event. In CIF data-based synthesis however, we currently use the linearize product CIF to CIF transformation to obtain the transtions (we call them edges). This means we may have multiple transitions per event. I think I'll first merge the edges per event into a single edge. That way, we have transitions at the same granularity as in the paper. We can always see later on if we can also allow different granularities for the workset algorithm. But, that requires careful consideration of the dependencies then. I think the per-edge vs per-event granularity can be implemented independently of using the workset algorithm or not. I'll create a separate issue for it, and solve that first. It could be useful in and of itself. It likely improves performance for some models, and not for others. It is likely a memory vs time trade-off.