workflows/eval: fail hard without target run

Without a target run, we won't get any rebuild labels, rebuild counts or
maintainer pings. This might have been correct before #373935, but by
now we run eval on all commits on the target branch, so we should treat
it as a failure if we can't find the run.

.github/workflows/eval.yml

@@ -175,7 +175,7 @@ jobs:
             --jq '.workflow_runs | sort_by(.run_started_at) | .[-1]') \
             || [[ -z "$run" ]]; then
             echo "Could not find an eval.yml workflow run for $BASE_SHA, cannot make comparison"
-            exit 0
+            exit 1
           fi
           echo "Comparing against $(jq .html_url <<< "$run")"
           runId=$(jq .id <<< "$run")
@@ -189,7 +189,7 @@ jobs:
 
           if [[ "$conclusion" != "success" ]]; then
             echo "Workflow was not successful (conclusion: $conclusion), cannot make comparison"
-            exit 0
+            exit 1
           fi
 
           echo "targetRunId=$runId" >> "$GITHUB_OUTPUT"