fix: guard optimize_refine_core against float32 noise oscillation

[yzhuang-els]

Fix #592

Modifications:

• add tol_optimization parameter to optimize_refine_core and require delta_local > tol_optimization for a move, instead of delta_local > 0
• thread Leiden.tol_optimization (default 1e-3, already exposed on the Python class) through to optimize_refine_core

Impacted submodules:

• sknetwork.clustering

This pull request:

• targets the develop branch
• does not decrease code coverage
• passes the tests
• is documented and the documentation build passes
• has PEP8-compliant code and explicit variable naming