Feature Request: Add Resume Functionality Support to Spark Plugin

### Feature description

## Question: Why doesn't the Spark plugin support resume functionality?

I would like to understand the design decision behind not supporting resume functionality in the Kestra Spark plugin, and request consideration for adding this feature.

## Problem Description

The current Spark plugin does not support job resumption, which causes data consistency issues in production scenarios:

### Scenario 1: Kestra Worker Restart/Deployment

When a Kestra Worker node restarts or deploys a new version while a Spark job is running:
- **Current Behavior**: After Worker restart, Kestra creates a brand new Spark task to retry
- **Problem**: If the user's Spark job uses `INSERT INTO` (append mode) instead of `INSERT OVERWRITE` (overwrite mode), the retry will cause duplicate data writes

## Request

Could you please clarify:
1. **Why resume functionality was not included** in the initial design?
2. **What are the technical or design considerations** that prevent or complicate adding resume support?

## Suggestion

We suggest adding resume functionality support to the Spark plugin. This would allow:
- Resuming Spark jobs when Worker restarts, instead of creating new tasks

## Priority

**High Priority** - This affects data consistency in production environments.

The issue names no files or tests. Start by reviewing the Spark plugin's job-submission lifecycle and Kestra Worker restart and retry behavior, especially for INSERT INTO jobs. Done would require an agreed resume design and evidence that a worker restart does not create a duplicate Spark submission.

Written by the indexing model from the issue text.