Different outputs for ncTestRun using float-typed inputs

[dschlaep]

Failed tests:
* testRun-19__dom-s-1-geog_in-s-geog-1-float
	** Check: ok -- expected outcome (success)
	** Comparison to reference: values: failed
	** Comparison to weather inputs: ok
	** Messages:
		-- 'AET_1980-1999_day.nc' is not equal to reference:
				 Component “evspsbl”: Mean relative difference: 0.0002742238,
				 Component “ec”: Mean relative difference: 0.0003555233,
				 Component “eow”: Mean relative difference: 0.002735918
		...	
		-- 'WETDAY_1980-1999_day.nc' is not equal to reference:
				 Component “wet”: Mean relative difference: 3

These differences arise because ​​on day 1992-52, the 4435-th day of the simulation run, average air temperature tas(float) is represented as 0.6100001335144043 which is just above the threshold of 0.61 for snowfall, whereas tas(double) was represented as 0.6099999999999999. Snowfall vs. no snowfall on that one day influenced infiltration, snowpack, snowmelt, surface water, etc. for about 60 days.