[ddl] Distributed ADD INDEX hangs in running/retry on persistent SetTSBeforeImportEngine context-deadline-exceeded errors

## Bug Report

> Evidence status: confirmed.

### 1. Minimal reproduce step (Required)

On testbed 8220955 with tidb_enable_dist_task=ON and tidb_ddl_enable_fast_reorg=ON, use a commit-matched failpoint owner and inject github.com/pingcap/tidb/pkg/ingestor/ingestctrl/mockAINativeSetTSBeforeImportEngineErr at SetTSBeforeImportEngine. One-shot control 1*return("context_deadline_exceeded") keeps job 4002 / task 300007 green. Persistent return("context_deadline_exceeded") keeps job 4007 and DXF task 300008 running for >90s; owner log records 247 repeated "meet retryable error" lines before clearing the failpoint, after which the same job syncs within about 2 seconds.

### 2. What did you expect to see? (Required)

A persistent retryable import/setup timeout should hit a bounded retry budget or escalate to a terminal failed/reverting state instead of leaving the distributed DDL running forever.

### 3. What did you see instead? (Required)

The DXF executor keeps context.DeadlineExceeded in the retry bucket, leaves the subtask running, and reruns it indefinitely; the job only finishes after external fault removal.

### 4. What is your TiDB version? (Required)

TiDB v9.0.0-beta.2.pre-1895-g5c9198e948 distributed ADD INDEX / DXF ingest path

Likely root cause and fix direction

DXF taskexecutor treats retryable subtask errors as stay-running with no per-subtask/per-DDL terminal retry budget, so a source-native retryable SetTSBeforeImportEngine timeout can bypass the lower ingest retry limit and become an unbounded rerun loop.