Add split size session config support in SparkScanBuilder.configureSplitPlanning()

[shanthoosh]

Problem

The session config spark.sql.iceberg.split-size was honored in some Spark read paths but ignored in others. Specifically, SparkReadConf.splitSizeOption() only checked the read option (SparkReadOptions.SPLIT_SIZE) and not the session config, causing inconsistent behavior: Specifically, APIs such as SparkStagedScan and SparkMicroBatchStream uses the session config(SparkSQLProperties.SPLIT_SIZE), while SparkScanBuilder.configureSplitPlanning() did not respect the session configuration(SparkSQLProperties.SPLIT_SIZE).

Fix

This PR fixes the inconsistency by updating splitSizeOption() to also consider the session configuration (SparkSQLProperties.SPLIT_SIZE), ensuring consistent split-size handling across all Spark reader control flows.

Testing

1. Added unit tests
2. Verified with ./gradlew clean && ./gradlew build

[sumedhsakdeo]

The name does mean that it is reading from Option alone.

[shanthoosh]

In SparkStagedScan and SparkMicroBatchStream, readConf.splitSize() method is used. Here is the relevant usage:

• https://github.com/linkedin/iceberg/blob/openhouse-1.5.2/spark/v3.4/spark/src/main/java/org/apache/iceberg/spark/source/SparkStagedScan.java#L46
• https://github.com/linkedin/iceberg/blob/openhouse-1.5.2/spark/v3.3/spark/src/main/java/org/apache/iceberg/spark/source/SparkMicroBatchStream.java#L103

[sumedhsakdeo]

We should take a deeper look at why only options / tableproperty are being looked at in the alternate split size path.

overriding method that is named to look at spark options alone with session conf / tblproperty, etc. seems incorrect.