Should [URL-level bounce tracker classification](https://github.com/privacycg/nav-tracking-mitigations/blob/main/explainers/bounce-tracking-mitigations-for-dual-use-sites.md#classification-of-bounce-tracker-candidate-urls) be done: 1. per client, 2. in aggregate, or 3. both? If aggregated classification of suspected bounce tracker URLs should be done, what thresholds should be used for classification?