The watermark table is in the data source because there is where you need to know what data changed in order to continue gather rows from where you left the previous time.
In case Azure Synapse is not the source but the destination, you can definitely create a table on Azure Synapse that can mirror the data source. Imagine the data source is Azure Table Storage, you can create a table on Synapse that mirror the Azure Table Storage, and where the watermark value could be the date field in the Azure Storage Table. You can create a pipeline in Azure Synapse that move the data and then run a stored procedure that recreate the MAX Date time.