Add the ability for judgements to be displayed as trends

[AJamesPhillips]

Currently

• Currently judgements are made given the current state of the system.
• It would be useful to see how the system has been changing over time. Like a mash up of a dashboard - complex system - causal map. Exactly like page 18 of https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/288846/10-632-land-use-futures-systems-maps.pdf

Description (acceptance criteria)

• [x] Allow a user to manually set the trend: "improving", "Little or no change", "deteriorating", "insufficient data / not assessed".
• [x] Write up some thoughts on this issue for how to calculate this automatically.

Why (hypothesis)

• Can see the trend over time across the whole system

Additional context

• Judgments are complex (trends even more so)
  1. Firstly judgements can change over time as the data changes. e.g. something might be bad if > 5 but later it's > 20 and that means it's then profitable enough to use for some other purpose that's more beneficial than its downside.
  2. Judgements can change over time even as the data stays the say due to our perceptions changing.
  3. the degree of a positive / negative judgement can change
• DataCurator allows for judgements to be "versioned" by setting an old judgement's validity to 0 and creating a new judgement.
• Calculating a trend from the data automatically is possible but there are many variables.
  1. Should the time window be the same for all trends in a particular knowledge view? e.g. in the linked to example, they are changes since 2003.
  2. Or should some trends be calculated differently?
  3. How is the "insufficient data" determined?
  4. Is there a moving average applied? If so how large?
  5. How is "Little or no change" or no change calculated? What defines "little"? a. Is it a degree of deviation from the existing data? b. Is it a degree of deviation from the existing and expected data? c. Is there a window applied to limit the time range of the data assessed? e.g. if there was a lot of volatility >10 years ago then you might like to just take the last 5 years of data to calculate if there was "Little" change.