feat(spark)!: Transpile ANY to EXISTS

[VaggelisD]

Fixes #4298

In Hive hierarchy ANY is an aggregate function for BOOLEAN expressions and not an array/subquery operator as is in other dialects, meaning that the following transpiled queries do not work:

>>> sqlglot.parse_one("WITH t AS (SELECT ARRAY[1, 2, 3] AS col) SELECT * FROM t WHERE 1 <= ANY(col)", dialect="postgres").sql("spark")
'WITH t AS (SELECT ARRAY(1, 2, 3) AS col) SELECT * FROM t WHERE 1 <= ANY(col)'

spark-sql (default)> WITH t AS (SELECT ARRAY(1, 2, 3) AS col) SELECT * FROM t WHERE 1 <= ANY(col);
[DATATYPE_MISMATCH.UNEXPECTED_INPUT_TYPE] Cannot resolve "any(col)" due to data type mismatch: Parameter 1 requires the "BOOLEAN" type, however "col" has the type "ARRAY<INT>".; line 1 pos 68;
...

However, Spark2+ supports the EXISTS() function which can process both ARRAY expressions and subqueries in a similar fashion. This PR:

1. Adds a transformation for the former (ARRAY) case, enabling the following path:

>>> sqlglot.parse_one("WITH t AS (SELECT ARRAY[1, 2, 3] AS col) SELECT * FROM t WHERE 1 <= ANY(col)", dialect="postgres").sql("spark")
'WITH t AS (SELECT ARRAY(1, 2, 3) AS col) SELECT * FROM t WHERE EXISTS(col, x -> 1 <= x)'

spark-sql (default)> WITH t AS (SELECT ARRAY(1, 2, 3) AS col) SELECT * FROM t WHERE EXISTS(col, x -> 1 <= x);
[1,2,3]

2. Extends EXISTS class to also inherit from Func, making it possible to parse & construct it as a proper function.

As step (2) is not required per se (could also build an anonymous function, check first commit), it's added as a standalone commit on top of (1); If we don't want to keep it it's trivial to drop it at once.

Docs

Postgres ANY for ARRAY | Postgres ANY for subqueries | Spark/Databricks ANY | Spark/Databricks EXISTS