Steve, in some places in the Mil (mainly training establishments) they use the graph as a go/nogo but I can't remember what limit they use. I seem to remember 2 mlux used to be a cutoff but then the Met Office changed their algorithms for calculating it and it became 0.7 which I think was the changeover to NVISlux.
In the SAR world we just use them as a guide since we don't, effectively, have any weather limits for SAROPs and for training have a cloud and vis limit rather than a light level limit.
As mentioned earlier, the problem with trying to forecast likely light levels is complicated by cultural lighting - it's really a case of suck it and see because even on the darkest of nights (in UK anyway) having NVG is still better than not, we just go lower and slower
Lower light levels just mean being more careful but generally they don't limit what we do.