How would shallow clone be more stressful for GitHub than a regular clone?

Shallow clones (and the resulting lack of shared history data) break many assumptions that packfile optimisations rely on.

See also: https://github.com/orgs/Homebrew/discussions/225