It's not a defect of the ECO classification, it's a mismatch of the "naming" being applied. The naming is not part of the ECO classification, we shouldn't blame ECO if the name doesn't quite match up.

1.d4 Nf6 2.c4 e6 3.g3 d5 4.Bg2 Be7 5.Nf3 O-O 6.O-O is at this point NEITHER open nor closed, that depends on what black does next. But looking at, for example, the chessgames.com ECO help page, it's already called "Closed Catalan" there. That's nothing to do with ECO. As kylemeister points out, in ECO 6...dxc4 transposes from E06 to E05, in the 1991 edition this is note 17.

As for why Ntirlis doesn't cover 4...Be7 5.Nf3 O-O 6.O-O dxc4, it's because he recommends 1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.g3 dxc4 5.Bg2 5...a6, which can't be reached from the 4...Be7 move order. I think Ntirlis's explanation of why he rejected 4...Be7 could have been clearer. Based on what he says about how he works, he probably considered 4...Be7 for the repertoire, rejected it for reasons, and then just didn't write down all the reasons in the book.

In [i]Playing 1 d4 d5[i] Ntirlis immediately dismisses the Closed " ...Be7 can arguably be considered the main line, but it gives White a few too many options...". All his subsequent discussions are of the Open Catalan. In the many games where early moves are given, Black always captures ...dxc4 on the 4th move.
In my database of major events past 12 years 46 % of Catalans are coded Closed E06 - E09 . But in 65 % of these ...dxc4 occurs later, usually on the 6th move. Other sources consider this as an Open Catalan , eg Kramnik vs Anand 2007 as given on the Wikipedia Catalan article. 
Is this just a fluke of an inadequate ECO coding ? Strange that Ntirlis doesn't mention & gives zero examples of the "delayed Open".