Helicopters are relatively inefficient flying machines when compared to fixed wing aircraft. In a hovering and low speed manouvering the engine is required to supply sufficient power to support the entire weight of the aircraft rather than being assisted by aerodynamic lift from winged surfaces. This means that weight is of critical importance in helicopter design. Every gram saved increases performance.
I highly suspect that the designers of the gearbox calculated the minimum size and strength requirements for the gear parts and added the necessary safety factors considering only the total cross sectional dimensions of the gears. This is perfectly satisfactory if the parts are of uniform material construction.
Problems arise when parts are surface hardened or carburised (carburized). The dimension of the parts are unaffected but an hardened and brittle outer layer is produced on the component which gives exceptional wear resistance. If for some reason the brittle hardened outer layer fractures, there may be no obvious signs of failure. The softer inner metal layer remains bonded to the hard outer surface and there may be no metal debris, chips or vibrational changes at all.
However, the load on that part is now being carried completely by the softer inner metal of that part. In effect the cross sectional area has been reduced somewhat and its load carrying characteristics are below what the original designer intended. The part does not fail immediately because of the safety factors designed in, but it is subjected to loads in excess of the design limits. The extra stress slowly causes the softer material to start to break up. The part may start to deform slightly and metallic debris and chips are most likely to be found at this stage.
Eventually the part will fail in overload. The problem is that the time span between a crack or fracture in the hardened surface and deformation and failure of the softer carrier metal cannot be calculated. It is unique for every event and dependent on too many variables to be calculated with any accuracy.
Such a damaged part may continue to function for years or fail suddenly due to a momentary extreme load such as an inadvertant abrupt control input causing a sudden demand for power, or sudden air turbulence imposing greater loads on the drivetrain.
So it may prove in this case that the design has shaved just a little too much off of certain critical parts in order to gain that extra bit of performance.
Aviation is by no means unique in having this type of failure mode. Motor racing has exactly the same problems. Designers keep stiffening components and reducing their weight and size until something breaks. Then it is redesigned and beefed up a bit to perform as it needs to. That is the way design progresses.