Enhance web app behavior when encountering snapshots with a large number of branches

The web app is not really usable on origins that yield snapshots containing a large number of branches.

(Example of an extreme case: https://github.com/v8/v8 has 110k branches)

Right now, the snapshot handling will try to cache objects for all branches, however many there is, which times out in extreme cases such as v8. Even just looking at the snapshot through the API times out because of the quantity of data to process.

We'll need some storage backend adaptations (e.g. to provide a paginated/filtered snapshot view), and some frontend adaptations to use these.

---

Migrated from T1207 (view on Phabricator)