This is a ES for the chip with pci-id 13E9, as stated by Apisak in his tweet from June 10, 2019:
I guess 13F8 may be the Dev Kit with defferrent PCI ID
There is also a new benchmark of Gonzalo.
tests in the same benchmark of ES 2G (13E9)
(Not 100%, but I think that's Gonzalo tests,even they hide the code)
"ES 2G" = 2nd Gen(Gen1) engineering sample of a gaming APU. Gonzalo with 13F8 pci-id was already in "Z" status, or QS by April 10, 2019. QS = essentially final version of chip. Chiphell and Komachi had been following 13E9 as far back as January, 2019.
---
I wouldn't focus too much on the HDD, as it's probably just part of the test bed they're using. As for the memory module sizes, yes, I expected different density based on the May 20, 2019 OQA PCB leak. It states, "16 Samsung K4ZAF325BM-HC18 in clamshell configuration". "325BM" is tied with 16Gb modules on Samsung's product selector, which would indicate 32GB of GDDR6 in clamshell, and you'd think 16GB in retail consisting of 8x 16Gb modules.
---
In all likelihood, it's as Apisak said, that the pci-id 13F8 with "Z" status is the dev kit, and if Reddit PCB leak is true(yet to be proven), then it was already being review by QA in May. I don't know enough about the minutiae of this process to say whether Flute is possibly using 16x 8Gb modules in clamshell, as opposed to half of the 16Gb modules of the alleged dev kit.
---
We know Gonzalo(13F8) went from ES1 to QS during the January to April time period. August 12, 2019 Komachi tweets concerning Oberon clocks and states, "For Oberon : ........../....GPU_ariel refs/.......". My guess is Oberon = retail chip.