Timebin optimization

This implements a change to the active/inactive checks following @jwillis' remark that the code was a bit slower than in the old master he was working on. 

The idea is to pre-compute the largest active time-bin and then use this information to decide whether act on particles or not. The old scheme was computing the end-time of the particle which involves a lot more information for the same result.

See merge request !316

Made all python scripts numpy 1.8.2 compatible, and fixed some bugs while doing it.

Title says it all. This should close #272.

See merge request !319

Cleaned up GIZMO code, added SineWavePotential tests.

The GIZMO code should now be clean enough to merge with `master`, and to be used in scaling tests.

I have added 1-3D versions of a new SineWavePotential test, which tests gravitational hydrostatic equilibrium in an unphysical but convenient 1D periodic sine wave potential. This allows testing gravity + GIZMO hydro without running into boundary condition problems. Unfortunately, the non conservative form of the potential makes it impossible to test energy conservation in these tests. GIZMO can keep all three set ups stable with both an isothermal and an ideal gas equation of state. I have currently only tested with the exact Riemann solver, but the approximate solvers should work as well.

The code is still not stable enough to cope with the hydrostatic and disc patch tests; these are next on my list.

See merge request !317

Removed ideal gas specific SineWavePotential scripts and incorporated them in the isothermal script. Ideal gas now seems to work.

Added 2D and 3D SineWavePotential tests. Need to run them at higher resolution and for longer, but they seem to produce reasonable results.

Added a maximal smoothing length that can be used to prevent particles from becoming too big in vacuum cases near the box edges. Will break the accuracy of the hydrodynamic scheme but is necessary in some situations.

Added some extra isothermal eos flags in Gizmo/hydro.h. Made sure gpart mass is updated when conserved mass changes. Fixed a number of possible issues with the isothermal Riemann solver.

Particle velocities are now correct. Period. Noh still does not work properly, but increasing resolution helps (so it is probably just a resolution issue).

Added some extra vacuum checks. Made sure isothermal Riemann solver crashes on vacuum (since isothermal vacuum cannot exist). Changed force.v_full to v instead of v_full and tried to explain why. Still have a problem with dt in acceleration computation which needs to be solved; will do this tomorrow.