Date columns showing incorrect integer values while doing CDC read

Date columns are showing integer values while doing CDC read.

For timestamp columns also it is converting into epoch , may be doing similar conversion for date column causing wrong integer values. 

Reproducible Code - 

```
columns = ["ts","uuid","rider","driver","fare","city", "report_date"]
data =[(1695159649087,"334e26e9-8355-45cc-97c6-c31daf0df330","rider-A","driver-K",19.10,"san_francisco", "2022-01-01"),
(1695091554788,"e96c4396-3fad-413a-a942-4cb36106d721","rider-B","driver-L",27.70 ,"san_francisco", "2022-01-01"),
(1695046462179,"9909a8b1-2d15-4d3d-8ec9-efc48c536a00","rider-C","driver-M",33.90 ,"san_francisco", "2022-01-01"),
(1695516137016,"e3cf430c-889d-4015-bc98-59bdce1e530c","rider-C","driver-N",34.15,"sao_paulo", "2022-01-01")]

inserts = spark.createDataFrame(data).toDF(*columns)

hudi_options =

{ 'hoodie.table.name': tableName, 'hoodie.datasource.write.recordkey.field' : 'uuid', 'hoodie.datasource.write.precombine.field' : 'ts', 'hoodie.datasource.write.partitionpath.field': 'city', 'hoodie.datasource.write.reconcile.schema':'true', 'hoodie.table.cdc.enabled':'true', }

inserts = inserts.withColumn("report_date", expr("CAST(report_date as timestamp)"))

inserts.write.format("hudi"). \
options(**hudi_options). \
mode("overwrite"). \
save(basePath)

cdc_read_options = \{ 'hoodie.datasource.query.incremental.format': 'cdc', 'hoodie.datasource.query.type': 'incremental', 'hoodie.datasource.read.begin.instanttime': 0 }

df = spark.read.format("hudi").load(basePath)
df.show(10, False)

spark.read.format("hudi"). \
options(**cdc_read_options). \
load(basePath).show(10, False)
```

## JIRA info

- Link: https://issues.apache.org/jira/browse/HUDI-7461
- Type: Bug
- Fix version(s):
- 0.16.0

Start with the supplied Spark reproduction and compare the regular Hudi read with the CDC read, focusing on the report_date and timestamp columns. Trace the CDC date and timestamp conversion path used by the incremental read. Done means date values remain readable dates and timestamp values are not incorrectly represented as epoch integers, with the reproduction covered by a regression test.

Written by the indexing model from the issue text.