[SPARK-29844][ML] Improper unpersist strategy in ml.recommendation.ASL.train

### What changes were proposed in this pull request?
Adjust improper unpersist timing on RDD.

### Why are the changes needed?
Improper unpersist timing will result in memory waste

### Does this PR introduce any user-facing change?
No

### How was this patch tested?
Manually

Closes #26469 from Icysandwich/SPARK-29844.

Authored-by: DongWang <cqwd123@gmail.com>
Signed-off-by: Sean Owen <sean.owen@databricks.com>

mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala

@@ -1046,13 +1046,13 @@ object ALS extends DefaultParamsReadable[ALS] with Logging {
       .persist(finalRDDStorageLevel)
     if (finalRDDStorageLevel != StorageLevel.NONE) {
       userIdAndFactors.count()
-      itemIdAndFactors.count()
-      itemFactors.unpersist()
       userInBlocks.unpersist()
       userOutBlocks.unpersist()
-      itemInBlocks.unpersist()
       itemOutBlocks.unpersist()
       blockRatings.unpersist()
+      itemIdAndFactors.count()
+      itemFactors.unpersist()
+      itemInBlocks.unpersist()
     }
     (userIdAndFactors, itemIdAndFactors)
   }