[runners-flink] Bump FlinkRequiresStableInputTest timeout to 120s (#38273)

The four portable-execution test methods in FlinkRequiresStableInputTest
were declared with `@Test(timeout = 30_000)`, which is consistently
exceeded under CI load and currently causes
testParDoRequiresStableInputPortable to fail
TestTimedOutException at FlinkRequiresStableInputTest.java:190 (the
1-second poll loop in executePipeline) on every recent
beam_PreCommit_Java run on master.

All four sibling portable-pipeline tests in the same package
(PortableExecutionTest, PortableTimersExecutionTest,
PortableStateExecutionTest, ReadSourcePortableTest) already use
`@Test(timeout = 120_000)`. This test is doing strictly more work
(deliberate failure -> checkpoint -> restore -> re-execute -> DONE) yet
had a 4x tighter budget, which matches the "Jenkins under high load"
diagnosis from #21333.

Aligning the timeout with the rest of the portable Flink test suite.

Author: tkaymak

runners/flink/src/test/java/org/apache/beam/runners/flink/FlinkRequiresStableInputTest.java

@@ -103,22 +103,22 @@ public static void setup() {
    * <p>A Savepoint is taken until the desired state in the operators has been reached. We then
    * restore the savepoint to check if we produce impotent results.
    */
-  @Test(timeout = 30_000)
+  @Test(timeout = 120_000)
   public void testParDoRequiresStableInput() throws Exception {
     runTest(false);
   }
 
-  @Test(timeout = 30_000)
+  @Test(timeout = 120_000)
   public void testParDoRequiresStableInputPortable() throws Exception {
     runTest(true);
   }
 
-  @Test(timeout = 30_000)
+  @Test(timeout = 120_000)
   public void testParDoRequiresStableInputStateful() throws Exception {
     testParDoRequiresStableInputStateful(false);
   }
 
-  @Test(timeout = 30_000)
+  @Test(timeout = 120_000)
   public void testParDoRequiresStableInputStatefulPortable() throws Exception {
     testParDoRequiresStableInputStateful(true);
   }