Modernization Tier 1: safe, mechanical Python cleanups

[shreyasgm]

Tier 1 of a codebase modernization pass. These are low-risk, mechanical changes with no behavior impact. They should all land in a single commit/PR.

Scope

• Fix broken install_requires in setup.py:21-23 — missing comma silently concatenates 'numpy >1.22.0' 'scikit-learn >1.0.0' into one string, so scikit-learn is never pinned. (Already flagged by users in Install problem and some question #32.)
• Remove unused imports
  • ComplexityData.py:5-8: sys, functools.wraps, time, datetime
  • calc_density.py, calc_proximity.py, coicog.py: unused pandas as pd
  • proximity.py:3-4: duplicate calc_proximity imports
• Replace == False / == True with boolean idioms
  • calc_proximity.py:22,25
  • ecomplexity.py:203,210,215
  • proximity.py:70,74,76
• Convert remaining .format() to f-string — ComplexityData.py:161
• Replace print() with logging / warnings — ecomplexity.py:103,176, proximity.py:52. Libraries shouldn't write to stdout unconditionally; verbose flag should control a logger.
• Drop class Foo(object) — ComplexityData.py:11. Python 3 implies object.
• Remove commented-out print calls in coicog.py:17,23.

Non-goals

No behavior changes, no API changes, no type hints, no packaging migration. Those are separate tiers.

Acceptance

• pip install -e . succeeds and correctly pins scikit-learn.
• All existing outputs identical (verify by running the README example and diffing against a saved reference if we can capture one — see the companion tests issue).