Improve diagnostics for datalog/NT queue capacity warnings

[PeterJohnson]

Both NT and DataLogManager have queues that are limited in size to prevent running out of memory. While a warning is printed when they reach capacity, it’s a fairly coarse warning and it is sometimes difficult to diagnose what in a user program is excessively logging. A common problem is the user logging an unbounded array, where a single log call can fill up a significant amount of the queue. Setting a lower threshold for individual log calls and overall providing more insight into what log calls are driving the top-level warning (eg largest 5-10 log entries in the queue with size+count) would be helpful diagnostic info to teams trying to identify the underlying cause.