Q1 optimizations

[szarnyasg]

Idea :bulb:

• do not compute induced subgraph beforehand
• perform an mxm to advance the wavefront
• then, check if the number of interactions of the edges was sufficient
• filter out those edges where it was insufficient

This might be faster than the current solution which precomputes the induced subgraph.