Avoid lazarus cancer/patients

[MClarke1991]

• We need to ensure that we discount paths in which the final score is good, but where at an intermediate stage the cancer score is so high that we assume the patient has a low chance of survival. I.e. treatments which take too long to have any effect on the cancer.
• We also need to account for the opposite, where the final score is bad but the cancer chose mutations which meant it would likely have died at an intermediate stage, and so is unrealistic. Although it would be good to be able to let the user choose whether to apply this assumption as they may want to assume that some cells survive even if they had Prolif = 0 Apop = Max.
• [ ] It would also be good to be able to keep some examples of these nonsensical paths so we can show them in a paper, maybe in the Supp, just to emphasise that we have accounted for this and are very smart.
• [ ] This can be solved by keeping the score at each level of the tree, as well as the final score at the very end.