Rather, the article describes the routes that compute views (e.g. show the change history across several commits) into history as being the issue. If we take that example, a JS client would have to fetch N individual commits/routes/objects and compute the requested view, but first it'd have to fetch the indexes/logs to determine what commits exist within e.g. a specified range.

I suspect that'd require more work on the frontend than "just run WASM-built git" ... unless the proposal is for it to fetch all requested objects lazily, in which case I think you'd be surprised how many files are read by git when answering a question like "show me the diff by user XYZ in file ABC on branch QRS between date 1 and date 2". That starts to get expensive to pull in the browser, and the bandwidth costs might start to hurt even if the backend now only had to serve cacheable dumb blobs.