Partially full refresh with microbatch

[nbenezri]

We have the following model in dbt-redshift:

{{ config(
    materialized='incremental',
    incremental_strategy='microbatch',
    full_refresh=false,
    unique_key='id',
    event_time='action_date',
    begin='2010-01-01',
    batch_size='day',
    lookback=7,
    sort='action_date', 
    dist='id'
) }}

SELECT
    s.id,
    s.cat_id,
    s.action_date,
    s.amount
FROM {{ ref('tab') }} s

This table holds billions of records from the last few years. Currently, cat_id only has one distinct value - 1. We are about to update the ref table to have cat_id: 2 with years of historical data. Of course, without full refresh this model will not have cat_id 2 with older data than this week. So I am thinking, how can I refresh old data by a specific column value without recreating the table?

I was thinking of adding something like:

{%- if var('cat_id_var', None) %}
        WHERE cat_id = {{ var('cat_id_var') }}
{%- endif %}

and call dbt with cat_id_var only when I need the full refresh, but, of course, doing full refresh will cause a table rebuild. Hope it is clear, would love any feedback/ideas.