For BigQuery we should consider copying primary key contraints

[nj1973]

Primary key constraints are not enforced therefore there should not be a performance overhead when inserting. BigQuery documentation states:

"
You should specify key constraints in the table schema when table data satisfies the data integrity requirements of primary key or foreign key constraints. The query engine can use the key constraints to optimize query plans.
"

Although there is a counter argument that we do not do this for any other SQL engine, it is left up to the customer to apply constraints. So perhaps this is overstepping the mark.

I'm logging this for discussion in the future.