Include BatchedSend state in cluster dumps

[gjoseph92]

BatchedSend is known to be brittle, and certain error cases can cause it to drop messages or even shut down entirely, causing deadlocks:

• Properly support restarting BatchedSend #5481
• Do not drop BatchedSend payload if worker reconnects #5457

I don't think we've seen these problems actually come up in the wild since #5525.

But when looking over cluster dumps, it would be nice to be able to rule this out as an issue. So I think we should include some minimal information about the state of BatchedSend, such as:

• len(buffer)
• please_stop
• waker.is_set()
• next_deadline

Obviously it would also be nice to include the status of the coroutine, but since we don't have a handle to it, that's lost to the ether of Tornado.