Hi,
using my own Tiger Grenadier I can confirm the rotten opening played with 30s. average/move.
With my usual setting (eval pawn 2 % / shield 25%) the Grenadier avoids the 3.Ba3 blunder (playing much better 3.Rb1) and, also one move before, plays a much correct 2.Nf3 instead of the rather poor move 2.Nc3
Shield seems to be the determining factor there, as minimal upgrade shield from 20 to 22% (while letting default eval pawn unchanged, so 0%) is enough to get the better moves I reported.
I think the engine is very sensitive, a tiny change in evaluation resulting in a much shuffled variation tree being selectively explored.
Thanks for the interesting games & comments!
MfG
Eric