Problem: two-phase drop takes a strong lock asynchronously and can cause transaction operations, which have 0-second lock request times, to fail randomly. Desired behavior: add majority writeConcern to such drop cmds to ensure the second phase finishes before testing proceeds; or the transaction op subsequent to the drop cmd needs majority write concern to achieve the same effect.

There needs to be some generic testing solution to two-phase drop interfering with transaction operations. It seems unlikely that every test writer will know to add majority writeConcern to drop cmds done in the proximity of transaction operations.