journal vdev truncation

[Besroy]

There is a concurrency issue where the leader performs both truncation and read operations simultaneously without lock: the m_data_start_offset and m_journal_chunks variables can be modified during a read operation. This results in inconsistencies where the read operation may fail to locate the expected chunk, ultimately leading to a system crash
More details in here