Catch time overflow during steady-state simulation

[dweindl]

• Remove the old and faulty check for finiteness of x in Model::fxdot. Closes Incorrect check for finiteness of x in fxdot #3076.
• The root cause of those NaNs in x was not CVODES. The problem is uncaught integer overflow during steady-state simulations and asking CVODES to integrate until t=inf. Catch that.
• Add a dedicated error code
• Add debug-level expectations to all integration functions of Solver

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.09%. Comparing base (47e2517) to head (faa42d2).
⚠️ Report is 34 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3077      +/-   ##
==========================================
- Coverage   77.73%   77.09%   -0.65%     
==========================================
  Files         313      313              
  Lines       20668    20670       +2     
  Branches     1499     1500       +1     
==========================================
- Hits        16067    15935     -132     
- Misses       4592     4726     +134     
  Partials        9        9              

Flag	Coverage Δ
cpp	71.18% <100.00%> (+<0.01%)	⬆️
cpp_python	37.24% <66.66%> (+<0.01%)	⬆️
petab	45.86% <0.00%> (-0.01%)	⬇️
petab_sciml	13.83% <0.00%> (-0.01%)	⬇️
python	69.83% <50.00%> (-0.02%)	⬇️
sbmlsuite-jax	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/amici.cpp	76.84% <ø> (ø)
src/forwardproblem.cpp	90.65% <100.00%> (+0.03%)	⬆️
src/solver.cpp	79.52% <100.00%> (+0.07%)	⬆️
src/solver_cvodes.cpp	72.25% <ø> (-0.15%)	⬇️

... and 5 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[FFroehlich]

👍

[FFroehlich]

Ah, it would actually be nice to stop simulation early (which the code so far did not achieve) and rather than simulating nonsense until maxsteps is reached.

[dweindl]

This isn't an issue anymore. Previously, model state became infinite only when asking CVODES to integrate until t=inf. This is now prevented at a higher level (EventHandlingSimulator::run_steady_state).