Bug: StandardTimeSeriesParameterValue don't calculate the write value for time-varying parameters

[mihlema]

The current SpineInterface access function for StandardTimeSeriesParameterValue does not adequately retrieve time-varying parameter values, as the averaging seems to assume equidistant values. The following example illustrates the problem:

julia> my_test_parameter = parameter_value(TimeSeries([DateTime(2000),DateTime(2001),DateTime(2008), DateTime(2009)],[5,5,0,0], false, false))
julia> my_test_parameter(t=TimeSlice(DateTime(2000),DateTime(2009)))
3.3333333333333335

In this example, we have a time-varying testparameter (distances: 1Y, 7Y,1Y). What we would expect when calling the value for the Timeslice(2000~>2009) would be the weighted average:

(5*1 + 5*7 + 5*0)/ 3

but we effectively get:

(5+5+0)/3

The source problem part is function (p::TimePatternParameterValue)(t::TimeSlice), where an unweighted mean is called

[mihlema]

I'd propose to just explicitly integrate the averaging in the function:
sum(p.values[a]*(p.indexes[a+1]-p.indexes[a]) for x = a:b)/sum(p.indexes[a+1]-p.indexes[a]) for x = a:b)