Yes, tuple concurrently updated because CREATE OR REPLCAE FUNCTION doesn't acquire an exclusive lock. You can see it with:

pgbench -c 10 -f /dev/stdin <<<"create or replace function test() returns int as 'select 42' language sql"