Debugging check failures with planetary simulations

Running a "typical" planetary simulation with --enable-debugging-checks leads to the following exits, immediately after [00000.3] engine_init_particles: Running initial fake time-step., where the next one in the list appears when the previous one is commented out:

• [00000.4] multipole.h:gravity_L2L():1956: Shifting tensors that did not interact
• [00000.4] multipole.h:gravity_field_tensors_add():299: Adding tensors that did not interact
• [00002.7] runner.c:runner_do_end_force():1802: g-particle (id=88808, type=Gas) did not interact gravitationally with all other gparts gp->num_interacted=88651, total_gparts=107900 (local num_gparts=107900)
• (After the first timestep) [00000.6] multipole.h:gravity_drift():244: Invalid maximal velocity

The last minimal/maximal velocity one is just a precision comparison issue because the error is raised even if velocity is actually equal.

The test case is a Uranus impact that has been run seemingly successfully with both SWIFT and HOT. Run here on cosma6 from the login node. On the planetary branch though atm that's identical to master. With swift/c5/intel/intelmpi/2017-parallel.

./configure --with-hydro=planetary --with-equation-of-state=planetary --with-tbbmalloc --with-multipole-order=4 --enable-debugging-checks

../../swiftsim/examples/swift -G -s -a -t 16 -n 2 pU_iU_m2L2v5_1e5.yml

A similarish second example throws [00000.7] multipole.h:gravity_drift():246: Invalid minimal velocity but no other issues.