Pushed a commit to apply parallel holds filtering within the main holds-to-target query. (My last TODO item from the code). With this, each parallel process only has to retrieve the holds it plans to process, instead of retrieving all targetable holds then filtering out the holds that should be ignored by the current parallel targeter slot.
Pushed a commit to apply parallel holds filtering within the main holds-to-target query. (My last TODO item from the code). With this, each parallel process only has to retrieve the holds it plans to process, instead of retrieving all targetable holds then filtering out the holds that should be ignored by the current parallel targeter slot.