Breakthrough time as an observation type

[berland]

Mismatch on breakthrough times (i.e. when some data vector crosses a given number in some direction, like when FWCT becomes more than 0.4) can be an important Observation type.

[berland]

Example yaml config could be (under general_observations):

  - name: Breakthrough times
    category: breakthroughtimes   # Even more generic name?
    observations:
      # Error unit is days. Trigger is the value of the summary vector that defines the
      # breakthrough.  Dates restrict the time interval for which the first date with
      # summary value above trigger is searched for, optional.  smooth-days will add a
      # rolling mean filter with the given number of days (use with care)
      - {key: FWCT, value: 2001-09-01, trigger: 0.4, error: 10, min_date: 2001-01-01, max_date: 2003-01-01}
      - {key: FWCT, value: 2000-02-01, trigger: 0.1, error: 3, min_date: 2000-01-01, max_date: 2003-01-01, smooth_days: 30}

[asnyv]

Even though breakthrough of water, gas or tracer maybe are the most obvious, I don't see a reason to limit the category name? Maybe "triggertimes" or something like that?
Should there be an option for which way the trigger works (goes above/goes below)? Or will the code be smart enough to "guess" that?
Also, if goes below is an option: probably need an exception for 0, which often means closed?