I believe that needs to be decided by the operator, taking into account their operation and fleets. For a jet on a short, normally wet runway, I would allow for a smaller margin than for a turboprop onto 4000m dry asphalt.
In the absence of any other information, maybe a sensible starting point could also be to look at what assumptions have gone into crew‘s landing performance calculations. On large jets, nowadays often a 7 second (or 1500ft) air distance is assumed between crossing the threshold and touchdown, in line with TALPA ARC recommendations. An air distance longer than that will invalidate the performance calculations (as this value is assumed to contain some margin already, using the full 1500ft will already use up all margin on the air part of the landing) and would thus provide a sensible event threshold. Obviously this would have to be reconciled against runway aiming point markings etc. in the FDM event validation to prevent a non-standard runway marking from generating false positives.