In the wtperf_run.py the list of stats to report for each test is determined by the -ops flag, or via a .json batch file (-bf flag).

We're currently processing all possible stats and then filtering them when they are printed, but it would be better to filter them up front when we build the StatConfiguration object.

Doing so will also allow us to print stats with a value of zero, which are currently filtered by the are_values_all_zero function. (This can be removed with this change).

When implementing this we need to check it works properly when arguments are passed in via a batch_file (-bf) as we will now need to define a new StatConfiguration object for each config present in the batch file.

Definition of Done:

• Stats to report are filtered when the StatConfiguration object is constructed
• Stats with a value of 0 are now reported by wtperf_run.py
• The framework correctly handles a batch file such as update-btree.json that contains multiple configurations