Trace a Memory Leak Across Repeated User Sessions

ROLE Act as a senior browser memory-debugging engineer. CONTEXT Application stack: [APPLICATION STACK] Repeated interaction: [USER INTERACTION] Memory evidence: [HEAP SNAPSHOTS OR METRICS] Relevant implementation: [CODE EXCERPTS] OBJECTIVE Determine why memory remains allocated after the interaction ends, and recommend the smallest defensible correction. INSTRUCTIONS Separate observed retention from normal caching and garbage-collection timing. Identify likely retention paths involving listeners, timers, subscriptions, detached DOM nodes, closures, or third-party instances. Rank hypotheses by evidence rather than familiarity. For each leading hypothesis, specify one experiment, its expected signal, and what would disprove it. Review ownership and cleanup responsibilities before suggesting changes. Provide a patch only where the supplied code supports one; otherwise request the exact missing excerpt. CONSTRAINTS Do not invent measurements, statistics, or profiler findings. Flag assumptions. Avoid recommending periodic page reloads as a fix. OUTPUT FORMAT Return an evidence summary, hypothesis table, focused patch or pseudocode, and repeatable before-and-after profiling procedure. QUALITY CHECK Explain how the proposed verification distinguishes a real improvement from delayed garbage collection.