Log-scaled data and log-scaled display shall be independent

As discussed with the SeaDAS development team, we shall make log-scaled displays (e.g. histogram/scatterplot axes, color point distributions) independent of the "logScaled" property of a RasterDataNode's source data. The "logScaled" property of a RasterDataNode's source data is used for compressing log-normally distributed floating point data into integer samples.

As a consequence, we'll have to make the statistics (which compute min,max,mean and histograms) work solely on geo-physical image data, not on the raw representation as it is now. --> Change implementation of org.esa.beam.framework.datamodel.Stx.

Then, we'll add log-scaled checkboxes everywhere where it makes sense:
- Colour manimuplation window, axis for the color sliders can be log-scaled (then also a different histogram must be shown)
- Statistics window, X-axis for the histogram can be log-scaled (then also a different histogram must be computed)
- Statistics window, X- and Y- axes for the scatter plot can be log-scaled