Add array_distance function

[austin362667]

Which issue does this PR close?

Partially closes #8782.

Rationale for this change

Add distance functionality to DataFusion. This function is particularly useful for scenarios involving spatial analysis, clustering, or similarity computations, where distance metrics are crucial.

It might be valuable to add scalar UDFs like list_distance/array_distance, similar to DuckDB, along with other methods of distance measurement (e.g., cosine etc)

XREF: DuckDB list functions, array functions

side note: In DuckDB, list and array have different definitions, but in DataFusion, they are considered the same.

What changes are included in this PR?

New function Euclidean array_distance(arr1, arr2) is added.

Are these changes tested?

Yes, added SQL logic tests.

Are there any user-facing changes?

New function array_distance(arr1, arr2) is added.

Examples:

> SELECT array_distance([1, 2, 3], [4, 5, 6]);
+-----------------------------------------------------------------------------------------------+
| array_distance(make_array(Int64(1),Int64(2),Int64(3)),make_array(Int64(4),Int64(5),Int64(6))) |
+-----------------------------------------------------------------------------------------------+
| 5.196152422706632                                                                             |
+-----------------------------------------------------------------------------------------------+
1 row(s) fetched.
Elapsed 0.010 seconds.

> CREATE TABLE points (
    point_a DOUBLE[],
    point_b DOUBLE[]
);
0 row(s) fetched.
Elapsed 0.039 seconds.

> INSERT INTO points VALUES
([1.0, 2.0, 3.0], [1.0, 2.0, 5.0]),
([2.0, 4.0, 6.0], [2.0, 4.0, 6.0]),
([1.5, 2.5, 3.5], [4.5, 6.5, 8.5]);
+-------+
| count |
+-------+
| 3     |
+-------+
1 row(s) fetched.
Elapsed 0.036 seconds.

> SELECT
    point_a,
    point_b,
    list_distance(point_a, point_b) AS euclidean_distance
FROM
    points;
+-----------------+-----------------+--------------------+
| point_a         | point_b         | euclidean_distance |
+-----------------+-----------------+--------------------+
| [1.0, 2.0, 3.0] | [1.0, 2.0, 5.0] | 2.0                |
| [2.0, 4.0, 6.0] | [2.0, 4.0, 6.0] | 0.0                |
| [1.5, 2.5, 3.5] | [4.5, 6.5, 8.5] | 7.0710678118654755 |
+-----------------+-----------------+--------------------+
3 row(s) fetched.
Elapsed 0.013 seconds.

No breaking change.

[Weijun-H]

> select distance(sq.column1, sq.column2) from (values (NULL, 2), (0,0)) as sq;
+---------------------------------+
| distance(sq.column1,sq.column2) |
+---------------------------------+
| 0.0                             |
+---------------------------------+

I prefer it to be NULL

[austin362667]

Sure let's do it this way~

[Weijun-H]

I am unsure how useful this feature would be because it only calculates two scalar column distances. 🤔

[jayzhan211]

@austin362667 I think array_distance is not aggregate function but array function. We tends to add new functions if other well-known DBs are also supported. In this case, I think it makes sense to have array_distance as array function but not aggregate function

[austin362667]

Got it! Thanks for reviewing.
Just adding list_distance in functions-nested and removing aggregate distance function from functions-aggregate.

[jayzhan211]

👍

We could support column value on the follow up PR

[Weijun-H]

Nice! Please add some doc here
## Array Functions

[Weijun-H]

Thanks, @austin362667, for the contribution, and @jayzhan211 for the review. We'll support the column value in a follow-up PR.