[CT-247] Support multiple unique_key for delete+insert incremental models (Postgres/Redshift/Snowflake/etc)

[jtcohen6]

I'd previously thought this would be harder than it is: #4618 (comment)

When unique_key is a list, all we need to do is adjust the delete statement:

dbt-core/core/dbt/include/global_project/macros/materializations/models/incremental/merge.sql

Lines 50 to 56 in 1df7a02

	{% if unique_key is not none %}
	delete from {{ target }}
	where ({{ unique_key }}) in (
	select ({{ unique_key }})
	from {{ source }}
	);
	{% endif %}

To use using, which should enable multiple unique_key match conditions. Something like:

delete from analytics.my_model a
using my_model__dbt_tmp b
where a.unique_column_1 = b.unique_column_1
  and a.unique_column_2 = b.unique_column_2
  ...

And a test for it, of course :)

[leahwicz]

@jtcohen6 is this for every adapter? Basically is that what "etc." meant?

[jtcohen6]

@leahwicz This change is relevant to any adapter that uses default__get_delete_insert_merge_sql. Postgres, Redshift, and Snowflake (specifically the delete+insert incremental strategy) are the only ones we need to be conscious of ourselves.

[jtcohen6]

resolved by #4858