I am starting to be paranoid for what Platon will start to be paranoid
No need to worry too much here, since the alert:007_ALERT_5_G Note: Number of Unrefined D-H Atoms ............ 4
is of level G (General Info/Check that it is not Something Unexpected) and type 5 (Informative Message, Check). In the explanations, PLATON suggests that "It is standard practice to refine Hydrogen atoms on hetero atoms as
proof of their correct assignment." And you are using constraints (AFIX 147 or AFIX 3) for the hydroxy hydrogen atoms, so that the O-H is not refined.
Since your data seems to be of good quality (given the small and nice ADPs, low s.u. of bond distances,...), and you are able to refine the water hydrogens with just a DFIX restraint and riding ADPs, I would suggest doing the same with the hydroxyl hydrogens (no AFIX, same DFIX restraint for O-H distance). But of course that is the way I would do it, and using AFIX is also fine.