Implicit window aggregation for cumulated sum

[Seddryck]

https://github.com/Seddryck/Tseesecake/issues/70 is a prerequisite.

This set of aggregation is a shortcut preventing the definition of Window. These aggregations automatically define a window where "PARTITION BY" is equal to "GROUP BY" and "ORDER BY" is equal to "BUCKET BY/Timestamp ASC".

If the SQL

SELECT
    Instant,
    Producer,
    SUM(Forecasted)
FROM
    WindEnergy
BUCKET BY 
    Year
GROUP BY
    Producer

is returning the following table

Instant	Producer	Forecasted
2022-01-01	Green Power Inc.	2737.176288
2023-01-01	Green Power Inc.	17782.837022
2022-01-01	Future Energy	1245.309402
2023-01-01	Future Energy	6739.813180

The following SQL

SELECT
    Instant,
    Producer,
    CUMULATED_SUM(Forecasted)
FROM
    WindEnergy
BUCKET BY 
    Year
GROUP BY
    Producer

Instant	Producer	Forecasted
2022-01-01	Green Power Inc.	2737.176288
2023-01-01	Green Power Inc.	20520.013310
2022-01-01	Future Energy	1245.309402
2023-01-01	Future Energy	7985.122582

The query translated to the dialect supported by DuckDB should be

SELECT
    DateTrunc('year', Instant) as Instant,
    Producer,
    SUM(SUM(Forecasted)) OVER (PARTITION BY Producer ORDER BY DateTrunc('year', Instant) ASC) as Forecasted  
FROM
    WindEnergy
GROUP BY 
    DateTrunc('year', Instant), producer