Make DiscusMultipleScatteringCorrection work with hollow\concave shapes

[DannyHindson]

The Monte Carlo simulation performed by DiscusMultipleScatteringCorrection doesn't work properly if the sample shape is either concave or hollow. The logic in DiscusMultipleScatteringCorrection::updateWeightAndPosition and DiscusMultipleScatteringCorrection::inc_xyz assumes there's only a single track segment inside the sample

This could be fixed by setting dl using Track::totalDistInsideObject instead of track.front().distInsideObject and then when incrementing the current position, the code could subtract off the track segment lengths until the current segment length is greater than dl

[RichardWaiteSTFC]

@warunawickramasingha and I think this already works actually with the current track generation - but we want to make a plot/simulation to show this and add to the documentation.