Refactor stats.py into focused modules

[tbrown122387]

Problem

src/ertimes/stats.py is 1,173 lines and contains 21 functions spanning unrelated concerns: ranking, capacity scoring, growth analysis, visualization, mental health analysis, duplicate detection, and more. This is a classic god-module problem.

The unresolved merge conflict (see #98) is a direct symptom — large, sprawling files are harder to review, harder to merge, and harder to keep correct.

Impact

High — The module is difficult to navigate, test in isolation, and extend without risk of unintended side effects.

Proposed Refactor

Split stats.py into focused modules:

New Module	Responsibility
stats_ranking.py	Hospital/county ranking functions
stats_analysis.py	Capacity scoring, mismatch detection, growth
stats_visualization.py	All plotting/mapping functions
stats_reports.py	Report generation, summaries

Keep a thin stats.py that re-exports the public API for backwards compatibility during the transition.

Acceptance Criteria

• No single module exceeds ~400 lines (rough approx. guess)
• All existing tests continue to pass
• Public imports from ertimes are unchanged

[Sarah-ko]

Handled in PRs #103 and #104