dbt_utils.unique_combination_of_columns test produces false failures in Redshift

[ClauPet]

Describe the bug

dbt_utils.unique_combination_of_columns test produces false failures testing for uniqueness of col_a and col_b in a model in Redshift consisting of only these two columns. While this error shows up when running the dbt_utils.unique_combination_of_columns test I get the same false failures when I run the complied test code directly in Redshift.

Steps to reproduce

I was unable to reproduce this error with an example model. I can only describe the scenario that I am seeing.

I have the following model:

          Column           |         Type          |   Collation    | Nullable | Default Value
---------------------------+-----------------------+----------------+----------+---------------
 col_a                     | character varying(20) | case_sensitive | YES      |
 col_b                     | character(4)          | case_sensitive | YES      |

I run the following test:

  - name: model_name
    data_tests:
      - dbt_utils.unique_combination_of_columns:
          combination_of_columns:
            - col_a
            - col_b

The compiled code looks like this:

select
      count(*) as failures,
      count(*) != 0 as should_warn,
      count(*) != 0 as should_error
    from (
with validation_errors as (
    select
        col_a, col_b
    from schema.model_name
    group by col_a, col_b
    having count(*) > 1
)
select *
from validation_errors
    ) dbt_internal_test

Expected results

I would expect no test errors. When checking the model for duplicates using sql or when using dbt core functionality to test for uniqueness I get no failing records.

The following test results in no failing records:

  - name: model_name
    data_tests:
      - unique:
          column_name: "(col_a|| '-' || col_b)"
```

### Actual results
I get 805 test errors, which would be the number of test errors if I had omitted col_b from the dbt_utils.unique_combination_of_columns and only tested for uniqueness of col_a

When I run the inner query of the compiled test code directly in Redshift:

```
with validation_errors as (
    select
        col_a, col_b
    from schema.model_name
    group by col_a, col_b
    having count(*) > 1
)
select *
from validation_errors
```

I get the following result:

```
 col_a | col_b
-------+------
(0 rows)
```
The result of the total query in Redshift is as follows:

```
 failures | should_warn | should_error
----------+-------------+--------------
      805 | t           | t
```


### System information
**The contents of your `packages.yml` file:**

**Which database are you using dbt with?**
- [ ] postgres
- [x] redshift
- [ ] bigquery
- [ ] snowflake
- [ ] other (specify: ____________)


**The output of `dbt --version`:**
```
Core:
  - installed: 1.8.8
  - latest:    1.9.1 - Update available!

  Your version of dbt-core is out of date!
  You can find instructions for upgrading here:
  https://docs.getdbt.com/docs/installation

Plugins:
  - postgres: 1.8.2 - Update available!
  - redshift: 1.8.1 - Update available!

  At least one plugin is out of date or incompatible with dbt-core.
  You can find instructions for upgrading here:
  https://docs.getdbt.com/docs/installation
```

### Are you interested in contributing the fix?
Depends on what issue this can be tracked down to