Image sharpness is not a linear scale

[adamkdean]

In the investigations into #276, we have discovered that with lovell/sharp library, sharpness processing doesn't work as expected. While previously with lwip it was a linear scale 1-100, it is now more advanced, and we need to improve the way that we perform sharpness.

The method we now use takes double const sigma, double const flat, double const jagged as parameters, so we need a way to map the linear scale 1-100 to these sigma, flat, and jagged variables.

[adamkdean]

🙄

[jimlambie]

@adamkdean @eduardoboucas to make this a bit simpler, I've modified the documentation so that it indicates the sharpness scale is 1-5. 5 looks awful...

Until we have some resource that can map the additional parameters to a linear scale, I think this will do.