Lake[].Epsilon is zero on day 1

https://github.com/AquaticEcoDynamics/GLM/blob/9fac22cbf9401ac08ee840c6239b6c992debd6aa/src/glm_flow.c#L248

Lake[Outflow_LayerNum].Epsilon is initialized to 0 for use in the outflow calculations and then set within deep_mixing in the subdaily loop. The outflow calculations in all subsequent days use the value updated by the subdaily loop (so not an issue for a multi-day run?). However, this means that a daily "hot start" always has a value of 0 for Epsilon. We need to move the code that sets Epsilon into the model initialization step. I think we should create a function in deep_mix.c called set_epsilon that can be called within glm_init.c. Thoughts @matthipsey?