stop processing if updated_deps is empty

[brettfo]

As part of the work inside Azure DevOps to improve dependabot (see #8179) there were other updates that needed to be split out into their own PR, specifically not continuing with an update if the updated set was empty.

[yeikel]

Under what situation can this happen? A test would help

[deivid-rodriguez]

I don't understand either why this is needed. If there's no update possible, we should be returning an :update_not_possible result from the UpdateChecker (a few lines above this).

If that's not happening, that sounds like a bug in the update checking process.

[brettfo]

We found an instance internally where the NuGet updater errored out and the result was an empty update set. While we fixed the specific issue we had, there still remains the possibility that another updater could also error out and result in an empty set, and that's why we ultimately added this check as well, to prevent the pr creator from crashing. If there is a better place for this check we can certainly move it.

[deivid-rodriguez]

I see.

I think it makes sense to avoid crashing randomly while trying to create a PR. However, I don't think this is an expected error condition, but a situation that's pointing at a Dependabot bug somewhere.

So, instead of just logging a message, we should be raising an error if this happens. Maybe something like:

raise "Dependabot found some dependency requirements to unlock, yet it failed to update any dependencies"

[ryanbrandenburg]

I have adjusted the change per your suggestion, added a test for this scenario, and updated the branch. The PR should be ready for re-review.