[Bug] The directory item limit of index is exceeded

### Search before asking

- [x] I searched in the [issues](https://github.com/apache/paimon/issues) and found nothing similar.

### Paimon version

paimon1.0.1

### Compute Engine

paimon-flink-1.19-1.0.1.jar

### Minimal reproduce step

I created a new table with pk like this:
caiji_time STRING,
dt STRING,
PRIMARY KEY (mid,dt) NOT ENFORCED
) PARTITIONED BY (dt) WITH (
'compaction.optimization-interval' = '7d',
'sink.committer-memory' = '500m',
'changelog-producer' = 'none', //No changelog file.
'deletion-vectors.enabled' = 'true', //MOW（Merge On Write）
'metastore.partitioned-table' = 'true',
'snapshot.num-retained.min' = '1',
'snapshot.num-retained.max' = '3',
'snapshot.time-retained' = '1h',
'metastore.partitioned-table' = 'true',
'file.format' = 'parquet',
'write-buffer-size' = '100mb',
'write-buffer-spillable' = 'true' ,
'partition.timestamp-formatter' = 'yyyy-MM-dd',
'partition.timestamp-pattern' = '$dt'
);

Then I customized a streaming job using KafkaSource and FlinkSinkBuilder. However, the files in the directory of index keep growing until my Hadoop reported: the item limit of /XXX/index is exceeded, they cause the task failed.

### What doesn't meet your expectations?

1. What are these index files?Why are they keep growing?
2. How can I resolve this problom?

### Anything else?

Nothing

### Are you willing to submit a PR?

- [ ] I'm willing to submit a PR!

No repository file or test is named; start with the reported KafkaSource and FlinkSinkBuilder reproduction and inspect how the Paimon index directory changes over time. Done means identifying what the index files represent, why they accumulate, and documenting or validating a resolution for the Hadoop item limit.

Written by the indexing model from the issue text.