Power Query has fewer rows than power bi datamodel? #postgreSQL

Sorry, but I have to come back with my problem:

Especially when it comes to duplicates (due to a buggy data connector...?) the solution cannot be to just use "Remove Duplicates". Because that means, that each and every row is loaded to memory - especially (!) when it comes to big datasets this cannot be the solution ?!

Here is my M-Code:

Query no1 (base query - not loaded):

############################

let
    Quelle = PostgreSQL.Database("localhost", "postgres", [CreateNavigationProperties=false]),
    public_costitems = Quelle{[Schema="public",Item="costitems"]}[Data]
in
    public_costitems

Query no 2 (loaded)

########

let
    Quelle = _costitems,
    #"Andere entfernte Spalten" = Table.SelectColumns(Quelle,{"id", "kind"}),
    #"Sortierte Zeilen" = Table.Sort(#"Andere entfernte Spalten",{{"id", Order.Ascending}})
in
    #"Sortierte Zeilen"

In Power BI I filter for an element with id=1495: the row is doubled. For no reason!

I have a guess - not really an explanation, but kind of a reason:

I logged all SQL-statements. I saw to sql-statements (pls ignore the quotation marks - I copied the statements out of the log files):

SELECT 
    ""$Ordered"".""id"", 
    ""$Ordered"".""itemnumber"", 
    ""$Ordered"".""overwrittenvalue"", 
    [...]
from ""public"".""costitems"" ""$Ordered"" 
order by ""$Ordered"".""id"" 
limit 4096"

And a second one which is nearly the same - onyl the limit/offset changed:

select

    [...]

from ""public"".""costitems"" ""$Ordered"" 
limit 9223372036854775807 offset 4096"

==>Why does it break the fold? Nothing special in the the query?

I looked at the datatypes of the columns of the tablle and I found, that there is a special datatype used for the column [kind] which is an "enumeration" in postgres. When I changed the selected columns and load e.g. the [orderid] instead of the column [kind], the rows and not doubled anymore. I can take an arbitrary set of columns of this table - as long as I do not use the field [kind] the query is folded and everything works fine. Only if I take the field [kind] in the selection, some rows are doubled.

But: I need this column. And it seems like a bug in the data connector. Anyone who can help me or give me a hint, where I can post the problem, if this forum is the wrong one?

Thanks

Holger