[CELEBORN-1720] Prevent stage re-run if another task attempt is running or successful
### What changes were proposed in this pull request?
Prevent stage re-run if another task attempt is running.
If a shuffle read task can not read the shuffle data and the task another attempt is running or successful, just throw the CelebornIOException instead of FetchFailureException.
The app will not failure before reach the task maxFailures.
<img width="1610" alt="image" src="https://github.com/user-attachments/assets/ffc6d80e-7c90-4729-adf7-6f8c46a8f226">
### Why are the changes needed?
I met below issue because I set the wrong parameters, I should set `spark.celeborn.data.io.connectTime=30s` but set the `spark.celeborn.data.io.connectionTime=30s`, and the Disk IO Utils was high at that time.
0. speculation is enabled
1. one task failed to fetch shuffle 0 in stage 5.
2. then it triggered the stage 0 re-run (stage 4)
3. then stage 5 retry, however, no task run in stage 5 (retry 1)
<img width="1212" alt="image" src="https://github.com/user-attachments/assets/555f36b0-0f0d-452d-af0b-1573601165e2">
4. because the speculation task succeeded, so no task in stage 5(retry 1)
<img width="1715" alt="image" src="https://github.com/user-attachments/assets/7f315149-1d5c-4c32-ae9b-87b099b3297f">
Due the stage re-run is heavy, so I wonder that, we should ignore the shuffle fetch failure, if there is another task attempt running.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
UT for the SparkUtils method only, due it is impossible to add UT for speculation.
d5da49d56d/core/src/main/scala/org/apache/spark/scheduler/TaskSchedulerImpl.scala (L236-L244)
<img width="867" alt="image" src="https://github.com/user-attachments/assets/f93bd14f-0f34-4c81-a8db-13be511405d9">
For local master, it would not start the speculationScheduler.
d5da49d56d/core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala (L322-L346)
<img width="1010" alt="image" src="https://github.com/user-attachments/assets/477729a4-2fc1-47e9-b128-522c6e2ceb48">
and it is also not allowed to launch speculative task on the same host.
Closes #2921 from turboFei/task_id.
Lead-authored-by: Wang, Fei <fwang12@ebay.com>
Co-authored-by: Fei Wang <cn.feiwang@gmail.com>
Signed-off-by: Shuang <lvshuang.xjs@alibaba-inc.com>
ad933815b6