[SUPPORT] Mismatched write token for parquet files

Use a Flink streaming job to write MoR tables. The compaction of a series of table was blocked by the following exception. It seems that the parquet file name in the compaction plan differs from the actual file name in terms of the write token part.

The actual file is `55078b57-488a-4be1-87ac-204548d3ec66_1-5-24_20230420023427524.parquet`.
```
2023-04-20 13:35:10 [pool-31-thread-1] ERROR org.apache.hudi.sink.compact.CompactOperator [] - Executor executes action [Execute compaction for instant 20230420041145422 from task 1] error
org.apache.hudi.exception.HoodieIOException: Failed to read footer for parquet s3a://path-to-table/dt=2023-01-20/hr=19/55078b57-488a-4be1-87ac-204548d3ec66_1-5-23_20230420023427524.parquet
at org.apache.hudi.common.util.ParquetUtils.readMetadata(ParquetUtils.java:95) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.common.util.ParquetUtils.readSchema(ParquetUtils.java:208) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.common.util.ParquetUtils.readAvroSchema(ParquetUtils.java:230) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.io.storage.HoodieAvroParquetReader.getSchema(HoodieAvroParquetReader.java:104) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.action.commit.HoodieMergeHelper.runMerge(HoodieMergeHelper.java:91) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.HoodieFlinkCopyOnWriteTable.handleUpdateInternal(HoodieFlinkCopyOnWriteTable.java:374) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.HoodieFlinkCopyOnWriteTable.handleUpdate(HoodieFlinkCopyOnWriteTable.java:365) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.action.compact.CompactionExecutionHelper.writeFileAndGetWriteStats(CompactionExecutionHelper.java:64) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.action.compact.HoodieCompactor.compact(HoodieCompactor.java:231) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.table.action.compact.HoodieCompactor.compact(HoodieCompactor.java:144) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.sink.compact.CompactOperator.doCompaction(CompactOperator.java:133) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.sink.compact.CompactOperator.lambda$processElement$0(CompactOperator.java:116) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.sink.utils.NonThrownExecutor.lambda$wrapAction$0(NonThrownExecutor.java:130) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source) [?:?]
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source) [?:?]
at java.lang.Thread.run(Unknown Source) [?:?]
Caused by: java.io.FileNotFoundException: No such file or directory: s3a://path-to-table/dt=2023-01-20/hr=19/55078b57-488a-4be1-87ac-204548d3ec66_1-5-23_20230420023427524.parquet
at org.apache.hadoop.fs.s3a.S3AFileSystem.s3GetFileStatus(S3AFileSystem.java:3866) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.s3a.S3AFileSystem.innerGetFileStatus(S3AFileSystem.java:3688) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.s3a.S3AFileSystem.lambda$getFileStatus$24(S3AFileSystem.java:3556) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.statistics.impl.IOStatisticsBinding.lambda$trackDurationOfOperation$5(IOStatisticsBinding.java:499) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.statistics.impl.IOStatisticsBinding.trackDuration(IOStatisticsBinding.java:444) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.s3a.S3AFileSystem.trackDurationAndSpan(S3AFileSystem.java:2337) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.s3a.S3AFileSystem.trackDurationAndSpan(S3AFileSystem.java:2356) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hadoop.fs.s3a.S3AFileSystem.getFileStatus(S3AFileSystem.java:3554) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at promoted.ai.org.apache.parquet.hadoop.util.HadoopInputFile.fromPath(HadoopInputFile.java:39) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at promoted.ai.org.apache.parquet.hadoop.ParquetFileReader.readFooter(ParquetFileReader.java:469) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at promoted.ai.org.apache.parquet.hadoop.ParquetFileReader.readFooter(ParquetFileReader.java:454) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
at org.apache.hudi.common.util.ParquetUtils.readMetadata(ParquetUtils.java:93) ~[blob_p-abdf98cc6fdb80521c5886e97d0250884f55321b-e6c0beee736c7301690a2ba078cc0a0f:?]
... 15 more
```

**Environment Description**

* Hudi version : bdb50ddccc9631317dfb06a06abc38cbd3714ce8

* Flink version : 1.16.1

* Storage (HDFS/S3/GCS..) : S3

* Running on Docker? (yes/no) :

**Additional context**

The job had metadata enabled first. I disabled the metadata table when restarting the job from a checkpoint.

Start with CompactionExecutionHelper.writeFileAndGetWriteStats, HoodieCompactor.compact, and the CompactOperator stack frames, then trace how the compaction plan obtains the parquet path. Compare the planned and actual filenames across the metadata-table-disabled restart scenario; done means the compaction no longer looks up a nonexistent write-token variant and the failure is covered by a reproducible test.

Written by the indexing model from the issue text.