Although simulators are excellent training tools there are some things that you need a real aircraft for
If what you claim (simulators do not have the required fidelity for crosswind landings) then this suggests the zero-zero flight time claimed by the manufacturers is bunkum and that regulators and airlines have been conned.
Some simulators do not have landing credits which is why pilots undergoing conversion are required by regulation to complete a certain number of take off and landings in the actual aircraft before being licenced to fly the type.
From your description of perceived poor qualities of crosswind landing fidelity in jet transport simulators, it would seem that pilots should have to complete a set number of crosswind landings in the actual aircraft before being endorsed on type. After all, the same principle applies to circling approaches where limitations to the visual depictions require pilots to demonstrate circling approaches in an aircraft - not necessarily the same type as the simulator. Admittedly it is not a flight fidelity issue but an outside view limitation where the airport cannot be seen over the shoulder.
From personal experience at strong crosswind landings real time in a 737 I have found the 737 Level D FFS I have operated have shown excellent fidelity in strong crosswinds.
The problem we often see in any full motion simulator with crosswind landings is not necessarily fidelity but a surprisingly number of pilots who are not only frightened of crosswind landings but simply do not know how to do them properly in the first place.
Hence the most frequent thing we observed is not only failure to round-out but failure to remove all drift before touch down. In turn this is caused by confusion on coordination of flying control input. This goes back to elementary flying school training where often inexperienced flying instructors themselves are frightened of crosswinds and avoid having to teach the exercise.