Let's evaluate each option to determine its accuracy:

• Option A is incorrect because checkpoints are used during the writing process, not reading.
• Option B is correct as each streaming query requires its own checkpoint path to ensure data integrity.
• Option C is correct because checkpoints support fault tolerance and data replay capabilities.
• Option D is the correct answer since it combines the truths of options B and C.
• Option E is incorrect due to the inaccuracy of option A.

For more details, refer to the documentation on Checkpointing in Spark Streaming.