Secret extraction via nonisolated redundant execution

Fault-injection weakness in PostQuantum-Feldman-VSS 0.8.0b2 and earlier allows physically proximate attackers to expose secret coefficients. The secure_redundant_execution function repeats operations in one Python process, compares results with timing-variable code, randomizes execution timing, and exposes partial-result timing through errors. Physical access to the host hardware is required; successful injection can disclose polynomial coefficients or make verification accept invalid shares or fraudulent commitments.