Add prior width option to use % of parameter value instead of float about pyautofit

[rhayes777]

Could I suggest that we use 'a' for absolute and 'r' for relative? It sits better with me than 'v' for value.

[rhayes777]

Also I think the relative expression should be a float. In this case 50% would be 0.5

[rhayes777]

This is now available as of 0.6.1. The 'a' marker will cause widths to behave as they did before. The 'r' marker will cause widths to be calculated by multiplying the associated value by the provided mean. For example, r, 0.5 with a mean of 2.0 would give a sigma of 1.0.

Relative widths should never be used to priors that could have a negative value.