Test if (open)hypergraphs are acyclic

[mstn]

Context

For rewrites in SMC we need acycle hypergraphs.

Details

• Added acyclicity checks for strict hypergraphs on the induced node graph; exposed is is_acyclic on Hypergraph and OpenHypergraph. We define the condition on strict because cycles can be the result of quotiening, if the graph is not strict, we might wrongly assume that it is acyclic.
• We defined node_adjacency similar to operation_adjacency where adjacency is IndexedCoproduct<K, FiniteFunction<K>>, intuitively a map nodes -> list of next nodes. We run kahn on node_adjacency for testing for cycles: Kahn’s algorithm removes all nodes if and only if the directed graph is acyclic (it runs on each connected components). Initial implementation used DFS, but better to reuse existing logic.
• Added tests: basic acyclic/cyclic cases, identity and symmetry acyclic, and a composed‑example that becomes cyclic.
• Refactored layer.rs, moved "graph" code in a separate module for reusability.

Known Issues and Tradeoffs

At the moment we don't cache or optimize memory, but it is something we might consider in the future.

• is_acyclic rebuilds node level adjacency each call; repeated checks will pay this cost repeatedly. We could cache adjacency.
• Full adjacency expands each hyperedge into source/target pairs, which can be memory‑heavy; a lighter node->outgoing‑edges index could reduce memory at the cost of per‑step traversal overhead.

[statusfailed]

only one minor change - operation_adjacency breaks public API. Please keep this one the same & add another function taking a Hypergraph argument if needed

[mstn]

• Made API pub(crate) as discussed on discord.
• Moved tests from integration tests to unit tests.