Division by zero in `update_edge_distance()` within `tsp_gls_2O`

[EricZheng1024]

In gls.py, the penalty matrix is initialized to zero within the _guided_local_search() and _guided_local_search_with_time() functions.

However, when the default update_edge_distance() function (located in template.py and evaluation.py) is called, this penalty (passed as edge_n_used) is used as a denominator:

combined_factor = (1 / edge_n_used) + (1 / edge_n_used)

Since NumPy is used for these operations, it triggers the following warning:
RuntimeWarning: divide by zero encountered in divide

Perhaps the default update_edge_distance() function should handle cases where edge_n_used is zero to avoid these division-by-zero warnings.