Hello Joao.
to me this looks like two issues: Noise or (more likely) no penetration of dense understory in the data (top picture). Chopped off mountain top (bottom picture).
The top picture is to be expected. You set the step so small that any small structures on the ground become visible and are not removed as you want an archaeological DTM. So in some cases you also retain low vegetation when it was very dense and not penetrable. A profile view through the point cloud should tell you for sure.
For the bottom picture I would be surprised if lasground were to "chop" off such a large mountain top. This can only happen for very large step sizes such as 20 or 50 and with your settings of step 0.8 it should be impossible. But it seems you have really run this with those settings as the "not penetrating dense low vegetation" also occurs. I assume that the mountaintop might be missing in the original point cloud? Maybe there was a cloud? Or severe low noise? Or some other odd thing? It would help (for the bottom issue) if you could post a link to that (LAZ) tile.
Regards from Panama,
Martin @rapidlasso