In designing PM-1825, we're planning to double the number of dhandles processed by adding in history store per table. In
WT-6421 I dove deep into how much time each piece of gathering dhandles took, both when we skip a clean table and when we need to apply the checkpoint to a table. That work drove a lot of PM-1407.
Now that the work for PM-1407 is in, reapply my timings of individual pieces to understand where dhandle processing is spending time and how much as research for PM-1825. Recall that the gathering dhandles phase of checkpoint is done holding the schema lock.