Update order_preservation.md

[soerenwolfers]

follow up to #4035 to explicitly answer misunderstandings exhibited in duckdb/duckdb#14755 and duckdb/duckdb#14941

By the way: Do you folks have recommendations for the common situation that one has a large table that's sorted by multiple columns (C1,...,Cn) and one wants to successively perform aggregations over Cn then Cn-1, ...? In that situation, there would be a trivial O(N) solution to doing each of the aggregations by hand, simply by checking group boundaries on the fly and producing aggregate values in an order preserving manner. Feels like the following duckdb snippet

SELECT  
  agg1(x)
FROM
(
  SELECT 
    agg2(x) AS x
  FROM 
  (
     ...
  )
  GROUP BY C2
)
GROUP BY C1


 would be inefficient in comparison?

[szarnyasg]

Thanks! I'll bring up the question with the team and get back to you.