Mark the replicas as replicas

Rails won't even attempt to write to a replica. Without this option,
a write attempt would actually run against Postgres, and it would
be up to Postgres to throw a readonly error.

Additionally, marking it as a replica teaches ActiveRecord that it
should not attempt to run migrations against this connection.

https://api.rubyonrails.org/classes/ActiveRecord/DatabaseConfigurations.html#method-i-configs_for

I'm actually pretty sure that the lack of `replica: true`
is why there's currently a db/slave1_schema.rb and db/slave2_schema.rb

config/database.yml

@@ -13,9 +13,11 @@ development:
   slave1:
     <<: *default
     url: <%= ENV['DB_SLAVE1_URL'] %>
+    replica: true
   slave2:
     <<: *default
     url: <%= ENV['DB_SLAVE2_URL'] %>
+    replica: true
 
 # Warning: The database defined as "test" will be erased and
 # re-generated from your development database when you run "rake".
@@ -43,6 +45,8 @@ production:
   slave1:
     <<: *default
     url: <%= ENV['DB_SLAVE1_URL'] %>
+    replica: true
   slave2:
     <<: *default
     url: <%= ENV['DB_SLAVE1_URL'] %>
+    replica: true