Stepping over events

[visr]

I found something peculiar when working on the test model for the Outlet:

In the current state it behaves as expected, that is: there is no outlet flow until the level boundary reaches the minimum crest level, and the basin level converges to the level boundary level. However, when the basin starts completely empty instead of with a tiny amount of water, it doesn't fill:

I haven't been able to figure out why this happens.

Originally posted by @SouthEndMusic in #549 (comment)

As can be seen from the dots on the blue line, the solver steps over the on-ramp entirely.

I modified the solver settings of the outlet test model, to use fixed time steps. This avoids the issue described above. It isn't clear to me why the initial state of 0 vs 0.001 m3 matters in this regard.

[solver]
adaptive=false
dt = 86400.0

We should understand this issue better, and see if we can do something about it. Right now we use tstops to tell the solver to stop when a flow boundary changes, but probably we need to include more moments here.