Don't get file contents from the DOM

Created by: ijsnow

We no longer want to retrieve file contents for extensions by scraping the DOM. This is because it's notoriously buggy, hard to maintain and comes with major performance problems when looking at large files or PRs with a lot of files.. This PR removes DOM scraping and uses Sourcegraph instances to fetch the file contents.

This means if you are using a Sourcegraph extension that requires file contents, the file must be in a repository on Sourcegraph.com or your own private Sourcegraph instance. @sqs, @ryan-blunden is there any particular way we'd like to communicate this to users? Note that extensions, like Codecov that don't need file contents still work as expected without a Sourcegraph instance containing private repos.

See this slack conversation for more context.

Closes https://github.com/sourcegraph/sourcegraph/issues/2000.