Decided it was unnecessary

* Winsorize means
* Only high outliers
* Scale edgeR dispersion

Fix bug that meant non-linear path factors weren't stored in output of
splatSimulatePaths

git-svn-id: https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/splatter@125902 bc3139a8-67e5-0310-9ffc-ced21a209358

* Add locfit to NAMESPACE
* Add return values to logistic, showPP documentation
* Delete commented mergeParams function
* Switch to aes_string in compareSCESets
* Add examples to getParam, setParam
* Adjust line lengths
* Adjust indentation

Getting note about scDatEx, but think it should be fine...

estimateSimpleParams -> simpleEstimate
estimateSplatParams -> splatEstimate

Neater than having a checkValid option