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

* Bump R version to 3.4
* Add row, col names to matrices

* 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

The last SCESets was been added multiple times
instead of adding each one onces.

Fixes #4

Make it clearer what is the package and what is the simulation

Validity method was set to SimpleParams

Not great, but will do for now

Zero means was causing NAs in counts

Some parameters weren't being passed from SCESet method to matrix
method

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