Price monitoring spec and implementation changes for parameters

[tamlyn10]

As per this Slack thread: https://vegaprotocol.slack.com/archives/C9E6XA8GK/p1603095646138900

And from a call with @barnabee and @witgaw we have decided:

• the list of α, τ, T (probability level, price projection horizon and auction extension duration) triplets be market parameters. This is implemented but needs spec clarity across price monitoring and market framework. In market framework, this will be a set of params on the market level (not tied to a trading mode).
• the parameter that specifies how often the risk model is called is going to be changed to be a network parameter (change in implementation and more clarity needed in spec). Going forward this may be a network parameter per risk model.

[edd]

A conversation I had with @campbellssource that may have come up previously is: while we're looking at these parameters, have we evaluated how they will need to change for 'the next product type' (e.g perpetuals)?

While supporting them immediately isn't on the rdevelopment oadmap yet, doing an exercise now of working out what the API would look like for those might lead us to make some API changes sooner rather than later.

[tamlyn10]

The price monitoring functionality is about how prices move and protecting against that, regardless of the product. What is interesting to think about is whether this functionality would work the same for different price determination methods. If/when we get to speccing a new one of those (such as frequent batch auctions) we will need to consider this.

[tamlyn10]

Our preference is that these parameters can default to levels set by a network parameter if unspecified. This will make education easier and allow market creators to avoid this complexity if they desire. Does that capture what you suggested the other day @barnabee ?

[tamlyn10]

Discussion and Barney's answer here... https://vegaprotocol.slack.com/archives/C9E6XA8GK/p1603288416316300?thread_ts=1603284418.295400&cid=C9E6XA8GK