Feature/identity asymmetry survey

[Lioo7]

Implement Identity Asymmetry Stress Test Strategy

Overview

This PR introduces the Identity Asymmetry Strategy, a new experimental survey condition designed to isolate "Project Identity Bias" from "Magnitude Bias." It allows researchers to determine if users favor specific budget categories (e.g., Health) over others (e.g., Education) when the mathematical utility of the allocations is identical.

Core Implementation

1. IdentityAsymmetryStrategy:
   • Identifies the largest pair of equal values in a user's ideal budget.
   • Generates a deterministic $n$-step stress test (default $n=10$) by shifting funds between the identified subjects.
   • Mathematical Integrity: To ensure the budget sum always remains exactly 100, the algorithm rounds the magnitude of the shift to the nearest integer before applying it to the vectors.
2. Dynamic Suitability Rules:
   • Added the min_equal_value_pair rule.
   • The system now validates that a user’s ideal budget contains a pair of identical values large enough to support the requested number of stress-test steps.

Analysis & Visualization

• Identity Consistency Score: Implemented a new metric in the Analysis Engine that calculates the strength of a user's bias (Max Wins / Total Questions).
• Aggregate Heatmap: The survey response dashboard now features a bucketed distribution table (50%–100%) to visualize bias intensity across the entire participant pool.
• The "Pain Curve": Individual user reports now include a visualization tracking the user’s choice at every magnitude step, identifying the "breaking point" where numerical loss finally overrides identity bias.

Testing

• Created tests/services/pair_generation/test_identity_asymmetry_strategy.py.
• Verified rounding logic, sum constraints, and metadata persistence.

Production Setup

To activate this survey in production, run the following SQL command:

INSERT IGNORE INTO surveys (story_code, active, suitability_rules, pair_generation_config)
VALUES (
    {story_code}, 
    True,
    '{"min_equal_value_pair": 10}', 
    '{"strategy": "identity_asymmetry", "params": {"num_pairs": 10}}'
);