[Improvement][Dependent Task] Optimize sub-workflow dependency logic

### Search before asking

- [x] I had searched in the [issues](https://github.com/apache/dolphinscheduler/issues?q=is%3Aissue) and found no similar feature requirement.

### Description

This is an extended issue for #16768 and #16776 to solve the ​scheduling time dependency​ issue.

The main problem is that dependent tasks query `WorkflowInstance` by `scheduleTime`, but sub-workflows do not have this value.

Recursively finding the main workflow is not a good approach, and iterating through all levels of sub-workflows from the main workflow is also inefficient—both may cause database performance issues.

Therefore, there should always be a ​key or value​ to directly determine the time and start action that workflows have.

For this reason, I propose creating a new object called `Action` to record the original workflow start information.

```mermaid
erDiagram
"[*Action]" {
int id
enum actionType
date submittedTime
}
"WorkflowInstance" {
int actionId FK
}
"TaskInstance" {
int workflowInstanceId FK
}

"WorkflowInstance" }|--|| "[*Action]" : ""
"TaskInstance" }|--|| "WorkflowInstance" : ""
```

Use `Action` addresses two issues:
1. Reducing database pressure: The dependent node first filters actions by submittedTime, then queries workflow-instances using (testFlag, code, actionId). This reduces table scans and optimizes index usage (previously on testFlag, code, scheduleTime/startTime).
2. Fixing dependency logic: Prevents null scheduleTime comparisons (in manual schedules), making the behavior more robust.

With the action object, the dependent node now query tasks always based on a fixed timestamp, not the node task's real `startTime`.

### Are you willing to submit a PR?

- [x] Yes I am willing to submit a PR!

### Code of Conduct

- [x] I agree to follow this project's [Code of Conduct](https://www.apache.org/foundation/policies/conduct)