[SPARK-53742][SQL] Push down the filter used in the count_if function

[r7raul1984]

What changes were proposed in this pull request?

By pushing down the filter condition in the count_if function, we can reduce the volume of data that needs to be processed.

spark.sql("create table t1(a int, b int, c int) using parquet")
spark.sql("select count_if(a <>1) from t1").explain("cost") 

Current:

== Optimized Logical Plan ==
Aggregate [count(if (NOT _common_expr_0#6) null else _common_expr_0#6) AS count_if((NOT (a = 1)))#4L], Statistics(sizeInBytes=16.0 B, rowCount=1)
+- Project [NOT (a#0 = 1) AS _common_expr_0#6], Statistics(sizeInBytes=1.0 B)
   +- Relation spark_catalog.default.t1[a#0,b#1,c#2] parquet, Statistics(sizeInBytes=0.0 B) 

After this pr:

== Optimized Logical Plan ==
Aggregate [count(if (NOT _common_expr_2#22) null else _common_expr_2#22) AS count_if((NOT (a = 1)))#21L], Statistics(sizeInBytes=16.0 B, rowCount=1)
+- Project [NOT (a#3 = 1) AS _common_expr_2#22], Statistics(sizeInBytes=1.0 B)
   +- Filter (isnotnull(a#3) AND NOT (a#3 = 1)), Statistics(sizeInBytes=1.0 B)
      +- Relation spark_catalog.default.t1[a#3,b#4,c#5] parquet, Statistics(sizeInBytes=0.0 B) 

Why are the changes needed?

Push down the conditional filter expressions in COUNT_IF to improve query performance.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Unit test.

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

No.

[beliefer]

Why not rewrite to SELECT count(_) FROM ut1 where c1 > 1 ?

[r7raul1984]

When there is only one count_if, it can indeed be rewritten as SELECT count(*) FROM ut1 WHERE c1 > 1. However, considering the consistency of the code logic, I personally feel that it's unnecessary to check the number of count_if statements.