Blocks fails to load reliably on slow networks

In a school setting with a less than ideal network, it very often takes students several reload attempts before code.world/blocks will load for them. Not sure if this is some kind of race condition, or just a consequence of the size of the generated JavaScript in the `funblocks-client` package.

Whatever the case, this is a serious problem. It not only takes 5 to 10 minutes away from each class session, but far worse, it leaves students feeling like can't use the site without adults there ready to handle the problems.

Reproduce the loading failure at code.world/blocks on a slow network and inspect the funblocks-client package and its generated JavaScript. Determine whether the behavior is a race condition or a payload-size problem, then define a reliable loading result that can be verified in repeated slow-network tests.

Written by the indexing model from the issue text.