[SPARK-55320][SQL][CONNECT] Use raise_error instead of divide by zero in Observation tests

### What changes were proposed in this pull request?

Use raise_error instead of divide by zero in Observation tests

### Why are the changes needed?

The tests fail when ANSI mode is disabled.

### Does this PR introduce _any_ user-facing change?

No

### How was this patch tested?

`SPARK_ANSI_SQL_MODE="false" build/sbt "sql/testOnly *DatasetSuite -- -z SPARK-55150"`
`SPARK_ANSI_SQL_MODE="false" build/sbt "connect-client-jvm/testOnly *ClientE2ETestSuite -- -z SPARK-55150"`

### Was this patch authored or co-authored using generative AI tooling?

Yes

Closes #54099 from heyihong/SPARK-55320.

Authored-by: Yihong He <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>

Author: heyihong

sql/connect/client/jvm/src/test/scala/org/apache/spark/sql/connect/ClientE2ETestSuite.scala

@@ -1620,7 +1620,10 @@ class ClientE2ETestSuite
     val observation = Observation("test_observation")
     val observed_df = spark
       .range(10)
-      .observe(observation, sum("id").as("sum_id"), (sum("id") / lit(0)).as("sum_id_div_by_zero"))
+      .observe(
+        observation,
+        sum("id").as("sum_id"),
+        raise_error(lit("test error")).as("raise_error"))
 
     observed_df.collect()
 

sql/core/src/test/scala/org/apache/spark/sql/DatasetSuite.scala

@@ -1164,7 +1164,7 @@ class DatasetSuite extends QueryTest
     val observed_df = spark.range(10).observe(
       observation,
       sum($"id").as("sum_id"),
-      (sum($"id") / lit(0)).as("sum_id_div_by_zero")
+      raise_error(lit("test error")).as("raise_error")
     )
 
     observed_df.collect()
@@ -1173,7 +1173,7 @@ class DatasetSuite extends QueryTest
       observation.get
     }
 
-    assert(exception.getCause.getMessage.contains("DIVIDE_BY_ZERO"))
+    assert(exception.getCause.getMessage.contains("test error"))
   }
 
   test("sample with replacement") {