That’s where 1.6% applies if I get it correctly.
You appear still to be missing the point that the WAT limit has nothing to do with obstacles and, vice versa, other than to the extent that, when running the NFP analysis, the floor is the WAT limit gradient for the particular segment being considered. If there are no obstacles of concern then, for the second segment (in which you appear to be interested) you will be limited either to the certificated MTOW (for which the gradient may well be greater than the WAT limit) or, if the MTOW gradient starts to compromise the 1.6% second segment WAT requirement, then you start to reduce weight to maintain not less than the WAT gradient requirement.
if an obstacle protrudes the net, then the net is elevated. Add 0.8% to that and you have the new gross gradient
Not quite. It is a mix and match between playing with weights and speeds to achieve the required net clearances. If, at your first guess weight, you aren't going to achieve the NFP clearance then you either reduce the weight and/or play with overspeed or use a curved departure. The NFP surface will increase in elevation (at that point) to get to the required clearance but the gradient isn't going to remain the same. Apart from the observation that knowing the gross gradient is useful to give you an idea of a minimum target ROC, gross gradient really is of little, if any, interest to us at all.