Cluster sources should always include all participating faults

[pmpowers-usgs]

Currently, as a site passes the 1000km distance cutoff in the CEUS (300km in the WUS), those faults in a cluster source that are beyond the cutoff drop out of the cluster calculation. It is preferable to always include all participating faults if any are in range. This also has implications for GMM weight changes at 500km; the participating faults in a cluster source may be computed using two different weights.

[pmpowers-usgs]

This issue only has an impact at long periods and in areas where rates of gridded seismicity are so low that far-field cluster sources contribute to hazard. However actual ground motion levels are usually so low (~0.01-0.05g) that the significance of this issue is limited.

Example of effect of total or partial inclusion of cluster source at 1000km cutoff (New Madrid is ~100km from Minneapolis):

[pmpowers-usgs]

nshmp-haz includes all or none of a cluster source; this will not be changed here.