Wrong values for MRCA priors after resuming a BDSKY run

[LuisEGomez]

I have a BDSKY analysis with several Uniform MRCA priors with the tipsonly option set to true. When I try to resume the analysis I get -Inf for the first of those priors (it contains a single taxa). After a little debugging I found that the starting value for that prior (1897.36) is not contained within the limits of the Uniform distribution (1898.0 - 1899.0). This "wrong" tip date is indeed present in the last sampled tree from the first MCMC, however, I also logged the tip dates and the same state has a correct value (1898.234) for the same sequence.

So I wonder why the .trees and the tipdates.log files have two different tip dates for the same sequence in the same state.