Counter point: spooky code at a distance is bad. Splitting your code to live partially in source control and partly in the database means keeping multiple layers in sync. This is coupling, and coupling multiple things, especially if that means teams, together means increased overhead.

I have seen business rules as stored procedures lock a business into their current model across with a dozen teams, effectively making system improvements impossible. And because they needed some olap atop oltp in some cases, their very beefy postgres solution crawled down to a max of 2k queries per second. I worked with them for over a year trying to pull apart domain boundaries and unlock teams from one another. Shared, stored procedures was a major pole in the tent of things making scaling the technical org incredibly hard.

Repeat after me: uncoupled teams are faster.