Support combined unique per store key

[indykoning]

This PR adds support for a combined uniquePerStoreKey
Not every database table has a single field which is duplicate with multiple stores, but unique otherwise.

An example of what this could work with before is the catalog_product_entity_media_gallery_value
value_id is completely unique except for store_id where it can have the same value_id

An example where this wouldn't work before is the catalog_product_entity_varchar table.
Where an entry is unique per store only when attribute_id and entity_id are combined. Since an attribute can have values attached and a product can also have multiple values attached, but a product cannot have multiple values of the same attribute attached.

An example of it's usage in that case would be:

new ForCurrentStoreWithoutLimitScope(['entity_id', 'attribute_id'], 'store_id')

[indykoning]

I've heavily optimised the query now, due to no longer using where not and concat it can take full advantage of the indexes. See the explain of a query heavily dependent on this feature:

id	select_type	table	partitions	type	possible_keys	key	key_len	ref	rows	filtered	Extra
1	PRIMARY	catalog_product_entity		ALL					29730	100.00	Using where
14	DEPENDENT SUBQUERY	catalog_product_entity_text		ref	CATALOG_PRODUCT_ENTITY_TEXT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_TEXT_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_TEXT_STORE_ID	CATALOG_PRODUCT_ENTITY_TEXT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	4	catalog_product_entity.entity_id	34	4.72	Using index condition; Using where
14	DEPENDENT SUBQUERY	eav_attribute		eq_ref	PRIMARY	PRIMARY	2	catalog_product_entity_text.attribute_id	1	10.00	Using where
16	DEPENDENT SUBQUERY	comparison		eq_ref	CATALOG_PRODUCT_ENTITY_TEXT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_TEXT_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_TEXT_STORE_ID	CATALOG_PRODUCT_ENTITY_TEXT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	8	catalog_product_entity_text.entity_id,catalog_product_entity_text.attribute_id,const	1	100.00	Using index
11	DEPENDENT SUBQUERY	catalog_product_entity_datetime		ref	CATALOG_PRODUCT_ENTITY_DATETIME_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_DATETIME_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_DATETIME_STORE_ID	CATALOG_PRODUCT_ENTITY_DATETIME_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	4	catalog_product_entity.entity_id	1	16.67	Using index condition; Using where
11	DEPENDENT SUBQUERY	eav_attribute		eq_ref	PRIMARY	PRIMARY	2	catalog_product_entity_datetime.attribute_id	1	10.00	Using where
13	DEPENDENT SUBQUERY	comparison		eq_ref	CATALOG_PRODUCT_ENTITY_DATETIME_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_DATETIME_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_DATETIME_STORE_ID	CATALOG_PRODUCT_ENTITY_DATETIME_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	8	catalog_product_entity_datetime.entity_id,catalog_product_entity_datetime.attribute_id,const	1	100.00	Using index
8	DEPENDENT SUBQUERY	catalog_product_entity_decimal		ref	CATALOG_PRODUCT_ENTITY_DECIMAL_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_DECIMAL_STORE_ID,CATALOG_PRODUCT_ENTITY_DECIMAL_ATTRIBUTE_ID	CATALOG_PRODUCT_ENTITY_DECIMAL_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	4	catalog_product_entity.entity_id	19	6.15	Using index condition; Using where
8	DEPENDENT SUBQUERY	eav_attribute		eq_ref	PRIMARY	PRIMARY	2	catalog_product_entity_decimal.attribute_id	1	10.00	Using where
10	DEPENDENT SUBQUERY	comparison		eq_ref	CATALOG_PRODUCT_ENTITY_DECIMAL_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_DECIMAL_STORE_ID,CATALOG_PRODUCT_ENTITY_DECIMAL_ATTRIBUTE_ID	CATALOG_PRODUCT_ENTITY_DECIMAL_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	8	catalog_product_entity_decimal.entity_id,catalog_product_entity_decimal.attribute_id,const	1	100.00	Using index
5	DEPENDENT SUBQUERY	catalog_product_entity_int		ref	CATALOG_PRODUCT_ENTITY_INT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_INT_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_INT_STORE_ID,CATALOG_PRODUCT_ENTITY_INT_ATTRIBUTE_ID_STORE_ID_VALUE	CATALOG_PRODUCT_ENTITY_INT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	4	catalog_product_entity.entity_id	25	6.86	Using index condition; Using where
5	DEPENDENT SUBQUERY	eav_attribute		eq_ref	PRIMARY	PRIMARY	2	catalog_product_entity_int.attribute_id	1	10.00	Using where
7	DEPENDENT SUBQUERY	comparison		eq_ref	CATALOG_PRODUCT_ENTITY_INT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_INT_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_INT_STORE_ID,CATALOG_PRODUCT_ENTITY_INT_ATTRIBUTE_ID_STORE_ID_VALUE	CATALOG_PRODUCT_ENTITY_INT_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	8	catalog_product_entity_int.entity_id,catalog_product_entity_int.attribute_id,const	1	100.00	Using index
2	DEPENDENT SUBQUERY	catalog_product_entity_varchar		ref	CATALOG_PRODUCT_ENTITY_VARCHAR_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_VARCHAR_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_VARCHAR_STORE_ID	CATALOG_PRODUCT_ENTITY_VARCHAR_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	4	catalog_product_entity.entity_id	43	6.83	Using index condition; Using where
2	DEPENDENT SUBQUERY	eav_attribute		eq_ref	PRIMARY	PRIMARY	2	catalog_product_entity_varchar.attribute_id	1	10.00	Using where
4	DEPENDENT SUBQUERY	comparison		eq_ref	CATALOG_PRODUCT_ENTITY_VARCHAR_ENTITY_ID_ATTRIBUTE_ID_STORE_ID,CATALOG_PRODUCT_ENTITY_VARCHAR_ATTRIBUTE_ID,CATALOG_PRODUCT_ENTITY_VARCHAR_STORE_ID	CATALOG_PRODUCT_ENTITY_VARCHAR_ENTITY_ID_ATTRIBUTE_ID_STORE_ID	8	catalog_product_entity_varchar.entity_id,catalog_product_entity_varchar.attribute_id,const	1	100.00	Using index

As you can see every subquery uses the index.
With a limit of 1, a query with this many tables to check takes 3ms
(with a dataset of 29k products, and in total 3.6mln product_entity_xxx records)

[royduin]

Do we need some unit tests for this so we're sure the results are the same?