## What changes were proposed in this pull request?

This was a suggestion by rxin at https://github.com/apache/spark/pull/17780#issuecomment-298073408

## How was this patch tested?

- modified existing unit test
- manual testing:

```
scala> hc.sql(" SELECT * FROM tejasp_bucketed_partitioned_1  where name = ''  ").explain(true)
== Parsed Logical Plan ==
'Project [*]
+- 'Filter ('name = )
   +- 'UnresolvedRelation `tejasp_bucketed_partitioned_1`

== Analyzed Logical Plan ==
user_id: bigint, name: string, ds: string
Project [user_id#24L, name#25, ds#26]
+- Filter (name#25 = )
   +- SubqueryAlias tejasp_bucketed_partitioned_1
      +- CatalogRelation `default`.`tejasp_bucketed_partitioned_1`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [user_id#24L, name#25], [ds#26]

== Optimized Logical Plan ==
Filter (isnotnull(name#25) && (name#25 = ))
+- CatalogRelation `default`.`tejasp_bucketed_partitioned_1`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [user_id#24L, name#25], [ds#26]

== Physical Plan ==
*Filter (isnotnull(name#25) && (name#25 = ))
+- HiveTableScan [user_id#24L, name#25, ds#26], CatalogRelation `default`.`tejasp_bucketed_partitioned_1`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [user_id#24L, name#25], [ds#26]
```

Author: Tejas Patil <tejasp@fb.com>

Closes #17806 from tejasapatil/add_serde.

(cherry picked from commit 814a61a867ded965433c944c90961df529ac83ab)
Signed-off-by: gatorsmile <gatorsmile@gmail.com>