Support for Arias intensity and Cumulative absolute velocity (CAV)

Very preliminary implementation AI and CAV within the nshmp hazard and disagg calcs. This will require some identification and implementation of GMMs and CGMMs for each intensity measure.

added Enhancement GMM labels

assigned to @amakdisi

changed the description

Prioritize conditional models? If yes, what model to use for each tectonic setting. Direct (?) models can be implemented as well, but separate NSHM GMM logic trees would be required; this is very easy to do and deploy to AWS for making map datasets.

Looking over our meeting notes from the fall, I think the initial plan would be to start with a conditional model implementation for Ia and CAV, and then eventually create a logic tree that uses both direct and conditional GMMs. I need to go back and recall which conditional GMMs are out there for which tectonic settings - we might have a gap in CEUS at the moment for conditional. Could instead use a direct GMM for that setting?

Here's a preliminary breakdown of which conditional GMMs could be used for crustal and subduction regimes (note: Sounds like Abrahamson et al. (2016) is being updated, I can talk to Jorge Macedo to see where that effort is at) - Farhadi and Pezeshk (2020) appears to be one of our only options for CEUS implementation, and are direct GMMs, rather than conditional.

IM	Crustal	Subduction	Stable Continental
Arias Intensity	Abrahamson et al. (2016) link to publication	Macedo et al. (2019) link to publication	Farhadi & Pezeshk (2020) link to publication
CAV	Macedo et al. (2021) link to publication	Liu and Macedo (2021) link to publication	Farhadi & Pezeshk (2020) link to publication

Perfect. There are better (more appropriate) conditional models for PGV that we've yet to get to but should be considered within this effort. We'll model the implementation off UsgsPgvSupport. There may be a better way to provide blanket conditional support if we know that no GMMs have AI or CAV.

Would going this route, at least early on, make things a bit more challenging further down the line if we opt for a blended approach (i.e. conditional+direct GMMs in the same logic tree)?

Are duration models of interest? (tangential to this issue)
e.g. https://doi.org/10.1785/0120240183

These were discussed back in 2021 at the summer User-Needs workshop, but pending a further lit review my instinct is to prioritize them a bit less due to the relative lack of coverage/epistemic uncertainty...