[SPARK-39534][PS] Series.argmax only needs single pass

### What changes were proposed in this pull request?
compute `Series.argmax ` with one pass

### Why are the changes needed?
existing implemation of `Series.argmax` needs two pass on the dataset, the first one is to compute the maximum value, and the second one is to get the index.
However, they can be computed on one pass.

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

### How was this patch tested?
existing UT

Closes apache#36927 from zhengruifeng/ps_series_argmax_opt.

Authored-by: Ruifeng Zheng <[email protected]>
Signed-off-by: Hyukjin Kwon <[email protected]>

Author: zhengruifeng

python/pyspark/pandas/series.py

@@ -6301,22 +6301,18 @@ def argmax(self, axis: Axis = None, skipna: bool = True) -> int:
         scol = scol_for(sdf, self._internal.data_spark_column_names[0])
 
         if skipna:
-            sdf = sdf.orderBy(scol.desc_nulls_last(), NATURAL_ORDER_COLUMN_NAME)
+            sdf = sdf.orderBy(scol.desc_nulls_last(), NATURAL_ORDER_COLUMN_NAME, seq_col_name)
         else:
-            sdf = sdf.orderBy(scol.desc_nulls_first(), NATURAL_ORDER_COLUMN_NAME)
+            sdf = sdf.orderBy(scol.desc_nulls_first(), NATURAL_ORDER_COLUMN_NAME, seq_col_name)
 
-        max_value = sdf.select(
-            F.first(scol),
-            F.first(NATURAL_ORDER_COLUMN_NAME),
-        ).head()
+        results = sdf.select(scol, seq_col_name).take(1)
 
-        if max_value[1] is None:
+        if len(results) == 0:
             raise ValueError("attempt to get argmax of an empty sequence")
-        elif max_value[0] is None:
-            return -1
-
-        # If the maximum is achieved in multiple locations, the first row position is returned.
-        return sdf.filter(scol == max_value[0]).head()[0]
+        else:
+            max_value = results[0]
+            # If the maximum is achieved in multiple locations, the first row position is returned.
+            return -1 if max_value[0] is None else max_value[1]
 
     def argmin(self) -> int:
         """