What do I collect before deleting an account?
The official export, a screenshot of the settings, and a note about what the export did not include. That last one is the part nobody writes down, and it is what you need when you open the archive years later and wonder what is missing.
What format survives a decade of neglect?
Plain text and HTML with no build step, images in formats older than the archive itself. The test is not whether it opens today but whether it opens with no software you have to install first.
Anything that changes this at ten times the scale?
At ten times the size the bottleneck moves from the operation itself to what it competes with: memory, locks and the people who have to run it. The approach stays, the batching gets smaller and the schedule matters more.