Issue during read from Redshift (via Spectrum) over a Hudi Table - AWS

### Bug Description

What happened:
We are facing a problem in a Data Platform based over AWS Cloud Infrastructure. After upgrading the EMR cluster from version 6.9 to 7.10, we are experiencing an issue with CDC-updated tables stored on S3. When reading the data through Redshift procedures, some records are not returned even though their Hudi commit time is earlier than the query execution time.

What you expected:
Some records with a Hudi commit time earlier than the query time should be correctly read and returned by Redshift (probably data under the same partition).

Steps to reproduce:

1. Upgrade the EMR cluster from version 6.9 to 7.10.
2. Run a PySpark script that implements CDC logic and writes data to S3 (Hudi tables).
3. While the PySpark CDC job is running (or after it completes), query the same data using Redshift procedures and observe that some records are not returned despite having a commit time earlier than the query time.

type of write operation in pyspark:
df_upsert_u_clean.write.format('org.apache.hudi') \
.option('hoodie.datasource.write.operation', 'upsert') \
.options(**hudiOptions_dfkkop) \
.mode('append') \
.save(S3_OUTPUT_PATH)

### Environment

**Hudi version:** 0.15.0-amzn-7
**Query engine:** PySpark
**Relevant configs:**
'hoodie.table.name': TABLENAME,
'hoodie.datasource.write.table.type': 'COPY_ON_WRITE',
'hoodie.datasource.write.recordkey.field': ','.join(key_cols),
'hoodie.datasource.write.keygenerator.class': 'org.apache.hudi.keygen.ComplexKeyGenerator',
'hoodie.datasource.write.partitionpath.field': 'PATH_KEY',
'hoodie.datasource.write.precombine.field': 'TSTMP',
'hoodie.index.type':'BLOOM',
'hoodie.simple.index.update.partition.path':'true',
'hoodie.datasource.write.hive_style_partitioning':'true',
'hoodie.datasource.hive_sync.enable': 'true',
'hoodie.datasource.hive_sync.table': TABLENAME,
'hoodie.datasource.hive_sync.database': 'ods',
'hoodie.datasource.hive_sync.partition_fields': 'PATH_KEY',
'hoodie.datasource.hive_sync.partition_extractor_class': 'org.apache.hudi.hive.MultiPartKeysValueExtractor',
'hoodie.datasource.hive_sync.use_jdbc': 'false',
'hoodie.datasource.hive_sync.mode': 'hms',
'hoodie.copyonwrite.record.size.estimate':3761,
'hoodie.parquet.small.file.limit': 104857600,
'hoodie.parquet.max.file.size': 120000000,
'hoodie.datasource.write.schema.evolution.enable': 'true'

### Logs and Stack Trace

_No response_

No repository files or tests are named. Start by reproducing the CDC upsert and Redshift read across EMR 6.9 and 7.10, then inspect Hudi commit and partition visibility for the missing records. Done requires identifying the compatibility or query-read cause and defining a verified fix or regression test.

Written by the indexing model from the issue text.