Resolve "Postgres timeout when counting number of CI builds for usage ping"

What does this MR do?

Instead of failing the whole usage ping when counts time out, replace them with -1 and continue. This means that the job shouldn't - 🤞 - retry, which will save us some queries in the long run.

From that data, we can create follow-up issues for individual teams to fix their own counts. I've already created https://gitlab.com/gitlab-org/gitlab-ce/issues/51097 and https://gitlab.com/gitlab-org/gitlab-ce/issues/51098.

For a sample ping, see https://version.gitlab.com/licenses/6257038.

Does this MR meet the acceptance criteria?

• Changelog entry added, if necessary
• Tests added for this feature/bug

Closes #45938 (closed).