Code for the critical re-evaluation of existing table union search benchmarks (Table Representation Learning workshop 2025), showing their limits for comparing methods.