Use dbt incremental model to update records based on a condition

[giacomochiarella]

I cannot find a way using dbt incremental model to update records based on a condition. Shortly, I would like to leverage the ability of incremental model to process only new data BUT I want the resulting records replace the existing ones only if column1 (in the resulting record) > column1 (in the existing record), of course based on a unique key.

Example:
Let's say my model extracts per order id the one having the greatest total
Very first run (full-refresh):
Input

order_id, total
1, 10
1, 20
2, 10
2, 20
3, 10

Output

order_id, total
1, 20
2, 20
3, 10

and all are added to the model given it is the very first run

Second run (incremental):
Input

order_id, total
1, 10
1, 15
2, 10
2, 15
3, 10
3, 15

Output

order_id, total
1, 15
2, 15
3, 15

But, given the same order ids exist with greater total except order_id = 3, only order_id = 3 must be replaced.
To implement this, I need to read the current data, but if I reference the model itself in the model dbt complains because it found a circular dependency. Can dbt do that?
I guess it could work if I could somehow instruct dbt to copy the model in a temporary table and reference this table to avoid the circular dependency. A bit hacky, would be possible in the pre_hook to execute a sql query to copy the current model in a temporary table in order to reference this model and avoid the circular dependency?