PING ZH.

Conflict with NLopt

5

`optimize` in Nonconvex will conflict with `optimize` in NLopt. Better resolve it when import both

Warn missing bound for models

1

Better first warn in `optimize` or somewhere else if haven't specify a bound for `Model`, there will be error.

Revert "Add cumulative distribution function"

3

Reverts noilreed/MultiKDE.jl#13

Better find a way to apply auto-derivative to sanity checks in test/kde.jl

Will be nice if we can choose to pre-initialize the whole KDE like [KernelDensity.jl](https://github.com/JuliaStats/KernelDensity.jl), using some optimized method like FFT or [AverageShiftedHistograms.jl](https://github.com/joshday/AverageShiftedHistograms.jl), which is needed in many situations.

Especially for categorical random variables

Will be good if add more kernels. An ideal way is link It to [Distributions.jl](https://github.com/JuliaStats/Distributions.jl) or [KernelFunctions.jl](https://github.com/JuliaGaussianProcesses/KernelFunctions.jl).

Find a better function signature for get_dict_candidates