[nll] Unweighted RMS computation

[luca-s]

Following the discussion in here we should consider if we want to compute the RMS in NLL plugin without weighting the residuals by arrival weight.

[luca-s]

Since changing the RMS computation might be consider a backward inconsistency we could consider to apply this change in the next release of SeisComP when the new NonLinLoc v7 code will land. That is already a big change

[gempa-jabe]

Will NonLinLoc 7 break catalogues by producing different solutions or different error measures? I am fine with changing the way the RMS is being computed, I just don't know how much it will affect current users.

Actually the name standardError suggests that the determination is kind of standard.

[luca-s]

Will NonLinLoc 7 break catalogues by producing different solutions or different error measures?

In my comparison tests between current NLL and NLL v7 I got similar results so I expect a transition to NonLinLoc 7 without any backward incompatibilities: neither in behaviour nor in configuration. NLL v7 should be a drop-in replacement to the current version. However I cannot be sure 100% since my testing is limited to my use cases.

I am fine with changing the way the RMS is being computed, I just don't know how much it will affect current users.

That would affect SED for example. So I pointed my collegues to this conversation. However for the sake of consistency I believe we should recompute RMS in a standard way.

[luca-s]

This has been discussed internally at SED and we decided it is not worth changing this metric and it would make future events inconsistent with historical ones.