New Diffs: Investigate infrastructure impact of caching server-rendered diffs

If we decide to go with the SSR approach to rendering diffs we might want to utilize Rails template caching. We need to calculate the approximate impact of this on our infrastracture (especially Redis), given then we'll be caching on a file by file basis (each file in MR gets a cache entry).

One way to approach that would be to calculate how many views we get per week (or day) in Gitlab SaaS for these pages:

1. Merge Request Diffs
2. Repository Commit details

We could use an approximate number of files in an MR (like 15) to calculate the potential number of Redis requests.

We'd also want to understand the number of visits to Compare revisions page, but since those are throwaway comparisons, we're likely less interested in caching those.