DRILL-735: Refactor HiveRecordReader
+ Create HiveFieldConverter classes for each Hive data type to extract value from Hive
ObjectInspectorand set it in Drill ValueVector. These objects are created in setup.
This way we avoid the switch case evalulation of type for each field in each row.
+ Use one form of Hive data type (TypeInfo) instead of two forms (string based type and type
from ObjectInspector). This simplifies the converter functions such as getMinorTypeFromHiveType,
convertPartitionType, etc.
+ Handle few primitive types that were not handled. Reading complex types is still a TODO.
+ Don't use HiveTextRecordReader as it is not completely implemented (few types are not handled)
+ Remove unused code: HiveInputReader.java
