Parameter initialization with ties and log-normal model

[flinder]

Theoretically ties are impossible under the log normal model because 0 diffusion times are outside the support of the distribution.
In the parameter initialization, there are many ties in the minimum diffusion times. Currently I'm ignoring them, but that leads to the lower bound being too high (although I guess it is unlikely that the target parameter is actually that low?) .
@bdesmarais do you have any suggestions on how to deal with this?