CatalogTableType constructor access modifier renders useless when case class factory is public

[pfcoperez]

What changes were proposed in this pull request?

CatalogTableType seems to implement an union data type enumerating all possible catalog table types.

It is built using a case class with a single value attribute "name". Wisely, it seems to be intended to be set only by the case class private constructor within the companion object with the idea of users not being able to create CatalogTableType instances but using those enumerated within that object.

However, making the case class constructor private does't forbid users creating whichever instances they want to create since the case class factory is still public. e.g:

scala> case class C private(x: Int)
defined class C

scala> C(1)
res0: C = C(1) // Should users be able to do this?

scala> new C(1) //This is the expected behaviour.
<console>:13: error: constructor C in class C cannot be accessed in object $iw
       new C(1)

This PR makes CatalogTableType a regular class, if matching the name was necessary, its companion object could implement unapply. Alternatives ways of solving this could be just using Scala's enumeration or model CatalogTableType using algebraic data types based on extending an interface, e.g:

trait CatalogTableType { val name: String } 

object CatalogTableType {
  object EXTERNAL extends CatalogTableType { val name = "EXTERNAL" }
  object INTERNAL extends CatalogTableType { val name = "INTERNAL" }
  object VIEW     extends CatalogTableType { val name = "VIEW" }
}

How was this patch tested?

This PR does not add new functionality and has thus been tested by your current test suite.

[AmplabJenkins]

Can one of the admins verify this patch?

[HyukjinKwon]

I guess this is meant to be internal API -

spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/package.scala

Lines 21 to 22 in f041e55

	* Catalyst is a library for manipulating relational query plans. All classes in catalyst are
	* considered an internal API to Spark SQL and are subject to change between minor releases.

Could you maybe open a JIRA and describe usecase to persuade make this a proper external API maybe?

Also, it'd be great if this PR follows http://spark.apache.org/contributing.html (e.g., [SPARK-XXXX] in the PR title with a JIRA).