[WIP] Add support of HiveTableScan TextHive

[amahussein]

Fixes #681

• look at the node description for the serde implementation as below

        org.apache.hadoop.hive.serde2.AbstractSerDe
          org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe -> HiveText
            org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe -> HiveParquet
            org.apache.hadoop.hive.serde2.avro.AvroSerDe -> AvroParquet
            org.apache.hadoop.hive.serde2.OpenCSVSerde -> CSVParquet

• Only HiveText is supported for now.
• we cannot pull the schema . So, we leave it empty

[amahussein]

@nartal1
We need to discuss in more details the changes as I found that there are some design issues that need to need addressed in order to fully support this feature.

1. In some situations, we need to check "Spark properties" in order to decide the action to take with a given operator/exec.
2. I believe that after adding this feature, DataSourceInfo is supposed to include the format of "Scan Hive/NativeScan" . Currently, this doe not happen because the tool looks for "ReadSchema" in the node info which does not apply for ScanHive. It will be nice to confirm if I am getting that right or not.
3. This point could be considered improvement; but it will be nice to generate a recommendation based on the spark properties. For example, if the CPU properties are not enabling Spark conversion, then we recommend that they do it. The point is where do we keep those recommendations or how do we pass it to the AutoTuner when it integrates with the Qual tool.

[nartal1]

2. I believe that after adding this feature, DataSourceInfo is supposed to include the format of "Scan Hive/NativeScan" . Currently, this doe not happen because the tool looks for "ReadSchema" in the node info which does not apply for ScanHive. It will be nice to confirm if I am getting that right or not.

You are right. Currently the tool looks for "ReadSchema" which is present in eventlogs for ORC and Parquet file formats. We try to get the schema as well. IIRC, schema is not present in the eventlogs for Scan Hive. We need to update that function to support Scan Hive file formats.

[nartal1]

Wonder if these are the only SerdeClasses that are used OR is there a possibility that a different class could be used for HiveParquet, HiveORC etc. Understand that we cannot cover all the classes, but would be good to document if that's the case.

[amahussein]

#723 supersedes this PR.