setting do_cic_particles = 1 and doPoissonSolve = 0 causes obscure error message

[chongchonghe]

Describe the bug
When we set do_cic_particles = 1 in a test problem (e.g., binary_orbit.cpp) and accidently set doPoissonSolve_ = 0 (the default), we get this error which is hard to debug:

0::Assertion `i < (this->std::vector<T, Allocator>::size())' failed, file "/Users/cche/softwares/quokka/quokka.copy3/extern/amrex/Src/Base/AMReX_Vector.H", line 35 !!!

To Reproduce
Steps to reproduce the behavior:

1. Compile this problem 'BinaryOrbitCIC' whiling setting sim.doPoissonSolve_ = 0;.
2. Run this problem
3. See error

Suggestion
This error occurs because the code tries to access phi in kickParticlesAllLevels() but phi is not initialized unless the Poisson solver is turned on. So, my suggestion is to abort the program if do_cic_particles = 1 && doPoissonSolve_ = 0. With the new particle interface, we should abort the program if we turn on any of the particle types with gravity and do not turn on Poisson solver.

Additional context
Add any other context about the problem here.

[markkrumholz]

Agree that that is a useful check to implement in the new particle architecture. Same thing for particle types that emit radiation and in simulations that are run with radiation turned off.

[BenWibking]

Yes, I agree. Feel free to add a PR with this check and an informative error message.

[BenWibking]

Since it doesn't actually cause wrong results or a failure in a case where it should work, I'm going to take off the red 'bug' label. But it should be more informative to the user, so I've instead added the yellow 'bug: usability' label.