[iceberg] Deprecate methods used for Iceberg support that break Parquet encapsulation

[parthchandra]

Some methods that are being used by Iceberg-Comet integration cause issues with Parquet shading. Alternate methods exist however they require Iceberg to be updated which in turn requires a Comet release.
To allow Iceberg test pipelines to succeed, we need to allow such methods until Iceberg can be updated and then deprecate them.

[parthchandra]

/comet/common/src/main/java/org/apache/comet/parquet/BatchReader.java

• public BatchReader(AbstractColumnReader[] columnReaders)
• public void setSparkSchema(StructType schema)
• public AbstractColumnReader[] getColumnReaders()

/comet/common/src/main/java/org/apache/comet/parquet/Utils.java

• public static ColumnReader getColumnReader( DataType type, ColumnDescriptor descriptor, CometSchemaImporter importer, int batchSize, boolean useDecimal128, boolean useLazyMaterialization)

/comet/common/src/main/java/org/apache/comet/parquet/TypeUtil.java

• public static ColumnDescriptor convertToParquet(StructField field)

/comet/common/src/main/java/org/apache/comet/parquet/ColumnReader.java

• void setPageReader(PageReader pageReader) throws IOException

/comet/common/src/main/java/org/apache/comet/parquet/ConstantColumnReader.java

• ConstantColumnReader(DataType type, ColumnDescriptor descriptor, Object value, boolean useDecimal128)

/comet/common/src/main/java/org/apache/comet/parquet/MetadataColumnReader.java

• MetadataColumnReader(DataType type, ColumnDescriptor descriptor, boolean useDecimal128, boolean isConstant)