A new invariant was added where operations that are holding open an oplog hole cannot try to acquire subsequent locks.

To temporarily circumvent this check, a new RAII-style class called IgnoreLockConstraintsOnTimestampedTxn was added to opt-out of the invariant.

There's a TODO in the codebase for this ticket. See SERVER-48129 for additional information.

The goal of this ticket should be to determine whether acquiring the lock can actually block in practice (perhaps another thread acquires a stronger lock).