Support hyperedges in TensorCircuit using cotengra with optimized execution.

This change introduces compatibility for hyperedges (CopyNodes) when using the cotengra contractor.
It features a new execution engine that avoids instantiating dense CopyNode tensors, preventing OOM errors on large hyperedges.

Key changes:
- `tensorcircuit/cons.py`:
    - Updated `_get_path_cache_friendly` to use UnionFind to group edges connected by CopyNodes.
    - Implemented a new primitive-based execution path (`_base`) using `einsum` on bare tensors. This handles hyperedges (shared indices) naturally without materializing large CopyNodes.
    - Preserved legacy contraction logic as a safe fallback when no hyperedges are present.
- Added `examples/hyperedge_demo.py` demonstrating the feature with a large-scale (20-leg) example.
- Added `tests/test_hyperedge.py` for verification using pytest fixtures.

Co-authored-by: refraction-ray <35157286+refraction-ray@users.noreply.github.com>

Author: google-labs-jules[bot]

tensorcircuit/cons.py

@@ -622,7 +622,7 @@ def _get_path_cache_friendly(
 
     size_dict = {}
     for root, symbol in mapping_dict.items():
-        size_dict[symbol] = root.dimension  # type: ignore
+        size_dict[symbol] = root.dimension  # type: ignore  # type: ignore  # type: ignore
 
     logger.debug("input_sets: %s" % input_sets)
     logger.debug("output_set: %s" % output_set)