I look forward to trying this out and seeing how this architecture change works in practice. For large Postgres databases (tens of TB in my experience), the old

I look forward to trying this out and seeing how this architecture change works in practice. For large Postgres databases (tens of TB in my experience), the old stats collector architecture was a reliable source of operational bugs that didn't have any real fixes, and this has been the case for a long time. If those issues start being addressed by these changes, that is a huge boon for people with large Postgres instances and will significantly improve its scalability story. This could be a really important change, especially for people with large instances.