[SPARK-21790][TESTS] Fix Docker-based Integration Test errors.

## What changes were proposed in this pull request?
[SPARK-17701](https://github.com/apache/spark/pull/18600/files#diff-b9f96d092fb3fea76bcf75e016799678L77) removed `metadata` function, this PR removed the Docker-based Integration module that has been relevant to `SparkPlan.metadata`.

## How was this patch tested?
manual tests

Author: Yuming Wang <wgyumg@gmail.com>

Closes #19000 from wangyum/SPARK-21709.

external/docker-integration-tests/pom.xml

@@ -80,6 +80,13 @@
       <type>test-jar</type>
       <scope>test</scope>
     </dependency>
+    <dependency>
+      <groupId>org.apache.spark</groupId>
+      <artifactId>spark-catalyst_${scala.binary.version}</artifactId>
+      <version>${project.version}</version>
+      <type>test-jar</type>
+      <scope>test</scope>
+    </dependency>
     <dependency>
       <groupId>org.apache.spark</groupId>
       <artifactId>spark-sql_${scala.binary.version}</artifactId>

external/docker-integration-tests/src/test/scala/org/apache/spark/sql/jdbc/OracleIntegrationSuite.scala

@@ -255,15 +255,6 @@ class OracleIntegrationSuite extends DockerJDBCIntegrationSuite with SharedSQLCo
     val df = dfRead.filter(dfRead.col("date_type").lt(dt))
       .filter(dfRead.col("timestamp_type").lt(ts))
 
-    val metadata = df.queryExecution.sparkPlan.metadata
-    // The "PushedFilters" part should be exist in Datafrome's
-    // physical plan and the existence of right literals in
-    // "PushedFilters" is used to prove that the predicates
-    // pushing down have been effective.
-    assert(metadata.get("PushedFilters").ne(None))
-    assert(metadata("PushedFilters").contains(dt.toString))
-    assert(metadata("PushedFilters").contains(ts.toString))
-
     val row = df.collect()(0)
     assert(row.getDate(0).equals(dateVal))
     assert(row.getTimestamp(1).equals(timestampVal))