Improve by argument in skill()

[jsmariegaard]

Expand by argument in skill() with two new features:

• "dt:" e.g. "dt:month" (which is different from "freq:month").
• "attrs:" e.g. "attrs:gtype" or "attrs:DA" (to distinguish between assimilation and validation stations)

Examples:

[ecomodeller]

Quarters is not the most natural aggregation period in natural systems. More relevant for business data.

How about aggregating by season?

Season	Abbreviation	Months
Winter	DJF	December, January, February
Spring	MAM	March, April, May
Summer	JJA	June, July, August
Autumn	SON	September, October, November

[jsmariegaard]

Quarters is not the most natural aggregation period in natural systems. More relevant for business data.

How about aggregating by season?

Season Abbreviation Months
Winter DJF December, January, February
Spring MAM March, April, May
Summer JJA June, July, August
Autumn SON September, October, November

Agree. It was not the best example. The idea with this feature is to provide access to the Pandas dt accessor on the index and be able to group by that. In Urban context I guess 'day_of_week' could be relevant, in natural environments it could be 'month'.

[jsmariegaard]

superseded by #351 and #352