[SPARK-53001][TESTS] Fix test_df_unpivot to pass with Spark 4.0

[sarutak]

Description

This PR proposes to fix the issue that test_df_unpivot doesn't pass with Spark 4.0.

---- dataframe::tests::test_df_unpivot stdout ----
SparkSession Setup

thread 'dataframe::tests::test_df_unpivot' panicked at crates/connect/src/dataframe.rs:2463:9:
assertion `left == right` failed
  left: RecordBatch { schema: Schema { fields: [Field { name: "id", data_type: Int64, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "var", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "val", data_type: Float32, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }], metadata: {} }, columns: [PrimitiveArray<Int64>
[
  1,
  1,
  2,
  2,
], StringArray
[
  "int",
  "float",
  "int",
  "float",
], PrimitiveArray<Float32>
[
  11.0,
  1.1,
  12.0,
  1.2,
]], row_count: 4 }
 right: RecordBatch { schema: Schema { fields: [Field { name: "id", data_type: Int64, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "var", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "val", data_type: Float64, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }], metadata: {} }, columns: [PrimitiveArray<Int64>
[
  1,
  1,
  2,
  2,
], StringArray
[
  "int",
  "float",
  "int",
  "float",
], PrimitiveArray<Float64>
[
  11.0,
  1.100000023841858,
  12.0,
  1.2000000476837158,
]], row_count: 4 }

As of Spark 4.0, ANSI mode is enabled by default but this test doesn't consider it.
To fix this issue, I tweaked the test so that the test pass in both ANSI mode is enabled or not.

Related Issue(s)

SPARK-53001

[xuanyuanking]

LGTM, will merge after we have the merge script!

[sarutak]

Now that we have a merge script am I OK to merge this by myself?

[xuanyuanking]

Yep, just go ahead for straightforward fixes! Thanks Sarutak!