Incorrect Histogram fit for arrays of equal Float64 values

[yurivish]

julia> using StatsBase

julia> fit(Histogram, [1 for _ in 1:1000])
Histogram{Int64,1,Tuple{StepRangeLen{Float64,Base.TwicePrecision{Float64},Base.TwicePrecision{Float64}}}}
edges:
  1.0:1.0:2.0
weights: [1000]
closed: left
isdensity: false

julia> fit(Histogram, [1.234 for _ in 1:1000])
Histogram{Int64,1,Tuple{StepRangeLen{Float64,Base.TwicePrecision{Float64},Base.TwicePrecision{Float64}}}}
edges:
  2.234:2.0:4.234
weights: [0]
closed: left
isdensity: false

Fitting a histogram to all-equal integer values works, but not for all-equal float values – see the 0 weight in the second example. The bin edges seem to be computed incorrectly?

[ararslan]

Fixed by #792