UI refactoring: correctly arrange Audit and Recommendations on the left column natively, Debate Arena on the right column

UI refactoring: separate Audit Specific Product into its own full-width top block to resolve CSS layout stacking issues and ensure 3 separate sections

UI refactoring: separate recommendation list into independent glass card, implement sticky debate arena, fix overflow clipping, and implement accordion view for simulated reviews

feat: embed dynamic simulated reviews directly in category recommendation cards

ux: remove auto-select on recommendation retrieval and filtering as per user instructions

ux: auto-select first recommendation and play debate immediately on fetch and filter to prevent blank placeholders

fix(tour): wrap Driver.js tours in robust try-catch and bump asset versions to v=2.0.2 for browser cache invalidation

feat: resolve silent button returns with UI alerts, ensure dual action path coexistence, and refactor heuristic review engine

Restrict LLM debate turns to exactly 6 single-round turns, and auto-select/sync first item on category filtering to prevent stuck details panels

Fix HF debate arena fallbacks, score focus 0 in LLM mode, and typewriter concurrent race condition

feat: implement custom product audit in Task B, enforce LLM re-ranking coverage, and fix missing price/category metadata in LLM recommendations

fix: resolve dimension alignment, cache-busting, and streaming early exit bugs

feat: persist active recommendation sub-tab, and dynamically synthesize Judge Agent debate verdict

feat: refine onboarding tour, unify textarea styling, update solution paper tabs, and resolve FastAPI 500 KeyError on custom personas

feat: enhance onboarding tour to walkthrough each tab and set groq as default model provider

feat: add driver.js onboarding tour for judges

feat: initial commit