[FEATURE] Introduce the lazy deletion mechanism for stage retry

### Code of Conduct

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

### Search before asking

- [X] I have searched in the [issues](https://github.com/apache/incubator-uniffle/issues?q=is%3Aissue) and found no similar issues.

### Describe the feature

Deleting the disk data costs too much time, let's introduce the lazy deletion for stage retry, it will be composed by 2 steps:
1. rename sync with register with the latest attempt number
2. putting the renamed path into the async deletion queue

### Motivation

_No response_

### Describe the solution

_No response_

### Additional context

_No response_

### Are you willing to submit PR?

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

The issue does not name files or tests. Start by tracing the stage-retry path that deletes disk data, then locate registration of the latest attempt and the asynchronous deletion queue. Done means renamed data is registered with the latest attempt and enqueued for asynchronous deletion without blocking retry.

Written by the indexing model from the issue text.