Regarding patterns it is specific to this component. I have since used the same data from CSV, where I converts it through tReplace, through double regexp's in a tMap, and have also used tAggregateRow
Now I tried to change the advanced setting for "Generation mode" from "Memory-consuming", to "Less memory consumed". That will also do the trick, and actually "Less memory consumed" reads all sheets faster than when using "Memory-consuming"
So I guess that formulas in the sheet and "Memory-consuming" do not work so well together.
And based on tests, it seems that "Less memory consumed" is faster anyway.
Memory-consuming (User mode): used for small file. It needs much memory. That is the reason