Added automatic data-quality checks and a real baseline-measurement tool

Two more pieces of measurement infrastructure: a daily automated check now looks for duplicate records across the platform's content databases and for mentor profiles missing a username (the exact kind of data gap that caused broken mentor links earlier in this project), catching this kind of thing automatically instead of waiting for someone to notice a broken link.

We also built the tool to compute real 30-day baselines for signups, verification, onboarding, and engagement, but we want to be straightforward about what it can and can't tell you today: event tracking only started recording data this week, so there is no 30 days of history yet to summarize. The tool is ready; the honest answer is that meaningful baseline numbers are still a few weeks out, and we'd rather say that plainly than manufacture a number that doesn't mean anything yet.

This closes out the measurement-baseline phase of our roadmap. The platform can now actually measure itself, not just watch for things breaking.