Support upload outputs and use needs context on Actions (#24230)

See [Defining outputs for
jobs](https://docs.github.com/en/actions/using-jobs/defining-outputs-for-jobs)
and [Example usage of the needs
context](https://docs.github.com/en/actions/learn-github-actions/contexts#example-usage-of-the-needs-context).

Related to:
- [actions-proto-def
#5](https://gitea.com/gitea/actions-proto-def/pulls/5)
- [act_runner #133](https://gitea.com/gitea/act_runner/pulls/133)

<details>
<summary>Tests & screenshots</summary>

Test workflow file:
```yaml
name: outputs
on: push

jobs:
  job1:
    runs-on: ubuntu-latest
    outputs:
      output1: ${{ steps.step1.outputs.output1 }}
      output2: ${{ steps.step2.outputs.output2 }}
    steps:
      - name: step1
        id: step1
        run: |
          date -Is > output1
          cat output1
          echo "output1=$(cat output1)" >> $GITHUB_OUTPUT
      - name: step2
        id: step2
        run: |
          cat /proc/sys/kernel/random/uuid > output2
          cat output2
          echo "output2=$(cat output2)" >> $GITHUB_OUTPUT
  job2:
    needs: job1
    runs-on: ubuntu-latest
    steps:
      - run: echo ${{ needs.job1.outputs.output1 }}
      - run: echo ${{ needs.job1.outputs.output2 }}
      - run: echo ${{ needs.job1.result }}
```

<img width="397" alt="image"
src="https://user-images.githubusercontent.com/9418365/233313322-903e7ebf-49a7-48e2-8c17-95a4581b3284.png">
<img width="385" alt="image"
src="https://user-images.githubusercontent.com/9418365/233313442-30909135-1711-4b78-a5c6-133fcc79f47c.png">



</details>

---------

Co-authored-by: Giteabot <teabot@gitea.io>

routers/api/actions/runner/utils.go

@@ -37,6 +37,17 @@ func pickTask(ctx context.Context, runner *actions_model.ActionRunner) (*runnerv
 		Context:         generateTaskContext(t),
 		Secrets:         getSecretsOfTask(ctx, t),
 	}
+
+	if needs, err := findTaskNeeds(ctx, t); err != nil {
+		log.Error("Cannot find needs for task %v: %v", t.ID, err)
+		// Go on with empty needs.
+		// If return error, the task will be wild, which means the runner will never get it when it has been assigned to the runner.
+		// In contrast, missing needs is less serious.
+		// And the task will fail and the runner will report the error in the logs.
+	} else {
+		task.Needs = needs
+	}
+
 	return task, true, nil
 }
 
@@ -124,3 +135,46 @@ func generateTaskContext(t *actions_model.ActionTask) *structpb.Struct {
 
 	return taskContext
 }
+
+func findTaskNeeds(ctx context.Context, task *actions_model.ActionTask) (map[string]*runnerv1.TaskNeed, error) {
+	if err := task.LoadAttributes(ctx); err != nil {
+		return nil, fmt.Errorf("LoadAttributes: %w", err)
+	}
+	if len(task.Job.Needs) == 0 {
+		return nil, nil
+	}
+	needs := map[string]struct{}{}
+	for _, v := range task.Job.Needs {
+		needs[v] = struct{}{}
+	}
+
+	jobs, _, err := actions_model.FindRunJobs(ctx, actions_model.FindRunJobOptions{RunID: task.Job.RunID})
+	if err != nil {
+		return nil, fmt.Errorf("FindRunJobs: %w", err)
+	}
+
+	ret := make(map[string]*runnerv1.TaskNeed, len(needs))
+	for _, job := range jobs {
+		if _, ok := needs[job.JobID]; !ok {
+			continue
+		}
+		if job.TaskID == 0 || !job.Status.IsDone() {
+			// it shouldn't happen, or the job has been rerun
+			continue
+		}
+		outputs := make(map[string]string)
+		got, err := actions_model.FindTaskOutputByTaskID(ctx, job.TaskID)
+		if err != nil {
+			return nil, fmt.Errorf("FindTaskOutputByTaskID: %w", err)
+		}
+		for _, v := range got {
+			outputs[v.OutputKey] = v.OutputValue
+		}
+		ret[job.JobID] = &runnerv1.TaskNeed{
+			Outputs: outputs,
+			Result:  runnerv1.Result(job.Status),
+		}
+	}
+
+	return ret, nil
+}