Add support for RGB images

DataLab provides image processing features for many data types but only grayscale.
Currently, here are the supported data types: `uint8`, `uint16`, `int16`, `float32` and `float64`.

This issue aims at proposing a new RGB data type which would be for example based on `uint8`.

This follows #118 reporting issues with image color spaces while trying to import a RGB image into DataLab.