RT normalization fails when chromosomes are too small

[dmalzl]

I currently used the pipeline and found it failing due to a chromosome that had too few values to smooth. An adhoc fix was to just skip loop iterations for such chromosomes. This should be incorporated

[dmalzl]

Specifically what I did was to simply bypass this behaviour by adding

if (length(RTb[,i+3]) < 10) next

to each loop. But it would be better to make this conditional a function of the binsize and LOESS span such that if the number of bins in the current array is smaller than the number of bins in the span, we skip the smoothing. This would then also make the exclusion of the M and Y chromosome obsolete and the whole process more general.