I tested it and got different value for F2. But I don’t know how to calculate that by hand:
Edit:
Oh… It depends on the offset of the reference point… yes it gives the same result.
Manual ccx 2.21. page 576.
“The location of the reference node is important since it is in this node that the resultant force is applied (this force may be defined by the user of may be the result of the calculation)”.
Additionally, Rotations imposed on the ROT are made arround the REF.
What about requesting SOF,SOM.? It requires much more refinement to be accurate but could provide the resultant forces and moments on the surface.
but the model with two supports shown not zero reactions by CalculiX also, maybe sketch a problem and hand calculation need to provide since beginning.
result are consistent with support definitions, below my random examples of similar cases.
I know that I used the wrong length for the lever arm in the second example, that wasn’t very smart… at least I quickly corrected it myself. 
Sorry, wasn’t my intention to point your comment directly. I should have reply as general, but it was your comment that made me remember that note on the manual.
Apart from that, hand calculations have well-defined pivot points to compute the level arms. The fixed hole doesn’t split equally the load FyA=FyB.
