better dealing with models that have stat3 when performing QP

[FaizLAB]

Please include a short description of enhancement here

I hereby confirm that I have:

• [ ] Tested my code on my own machine
• [ ] Followed the guidelines in the Contributing Guide
• [ ] Selected develop as a target branch (top left drop-down menu)

(Note: You may replace [ ] with [X] to check the box)

[rmtfleming]

maxDeviation param.rounding

need io commenting as per style guide: https://opencobra.github.io/cobratoolbox/stable/contributing.html#style-guide

Note that checking for optimality is better to be done at a lower level because it applies to more problems: https://github.com/opencobra/cobratoolbox/blob/master/src/base/solvers/solveCobraLP.m#L1598 https://github.com/opencobra/cobratoolbox/blob/master/src/base/solvers/solveCobraQP.m#L1030

Re-running solveCobraQP or solveCobraLP on a rescaled model is better done in optimiseCbModel for similar reasons.