[SUPPORT] PySpark reading hudi partition column for hudi table incorrectly

**_Tips before filing an issue_**

- Have you gone through our [FAQs](https://hudi.apache.org/learn/faq/)?

- Join the mailing list to engage in conversations and get faster support at dev-subscribe@hudi.apache.org.

- If you have triaged this as a bug, then file an [issue](https://issues.apache.org/jira/projects/HUDI/issues) directly.

**Describe the problem you faced**

A clear and concise description of the problem.

We have a hudi table with the following relevant `hudi.properties`:
```
hoodie.table.precombine.field=updated_at
hoodie.datasource.write.drop.partition.columns=false
hoodie.table.partition.fields=inserted_at
hoodie.table.type=COPY_ON_WRITE
hoodie.archivelog.folder=archived
hoodie.table.cdc.enabled=false
hoodie.populate.meta.fields=true
hoodie.partition.metafile.use.base.format=false
hoodie.table.version=5
hoodie.timeline.layout.version=1
hoodie.table.base.file.format=PARQUET
hoodie.table.recordkey.fields=id
hoodie.table.keygenerator.class=org.apache.hudi.keygen.TimestampBasedKeyGenerator
```
It may be worth noting that our deltastreamer is passed:
```
hoodie.deltastreamer.keygen.timebased.output.dateformat='yyyy/MM/dd'
```
as configuration.

and the `inserted_at` column is a `LongType`. The avro schema from an example parquet file is below:
```json
{"name":"inserted_at","type":["null","long"],"default":null}
```

When reading this table using the following `pyspark` code, the `inserted_at` column, even when `inferSchema:false` and the schema is explicitly set when loading the pyspark dataframe, displays the hudi output format of `yyyy/MM/dd` instead of the `LongType`, even when casting.

```python3
base_path = f"s3a://{bucket_name}/{table_prefix}"
df = (
spark_session.read.options(**hudi_opts)
.format("hudi")
.load(
base_path,
schema=T.StructType(
fields=[
T.StructField("inserted_at", T.LongType(), nullable=True),
# other columns
]
),
)
)
df.show()
```

**To Reproduce**

Steps to reproduce the behavior:

1. write hudi table with similar configuration
2. read from table pyspark

**Expected behavior**

A clear and concise description of what you expected to happen.

- I'd expect to be able to read the `LongType` value correctly

**Environment Description**

* Hudi version : 0.13

* Spark version : 3.1

* Hive version :

* Hadoop version :

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

* Running on Docker? (yes/no) : No

**Additional context**

Add any other context about the problem here.

**Stacktrace**

```Add the stacktrace of the error.```

Start by reproducing the PySpark read with the provided Hudi table properties, LongType schema, Spark 3.1, and S3 storage. Compare the resulting inserted_at schema and values with the Parquet/Avro schema and determine which Hudi or Spark entry point applies the yyyy/MM/dd partition format; done means LongType values are read correctly.

Written by the indexing model from the issue text.