Use consistent form in inequality constraint equations

[jonmaddock]

Following an investigation into upper limit constraint behaviour, I'm proposing to change the form of the upper limit inequality constraints from:
$$c_{max} = 1 - \frac{x_{max}}{x}$$
to
$$c_{max} = \frac{x}{x_{max}} - 1$$
This means that lower and upper limit constraints are violated proportionately (e.g. linearly), and the constraint values for upper limit constraints are now much simpler to interpret. It also removes a discontinuity in upper limit constraints when the constrained parameter ($x$) passes through 0.