Moved the un-skipping of tasks to a separate function. Abort drifting when forcerbuild is on. Access forcerebuild atomically.

Add debugging checks to make sure all the active tasks are scheduled and that no inactive task is being run.

Test gdwarf support in GCC compiler

Fixes a problem introduced by !261.

See merge request !262

Add compiler flags that should show inlined functions when enabling debug

Should work with Intel ICC and GCC.

See merge request !261

Tom feedback

Adds a new task, which depends on kick, to perform feedback. Currently, increases the thermal energy of the particle nearest a specified SN location by a specified amount.

See merge request !232

Sanitize smoothing lengths

Here is a proposal to address problems such as #211. 

At the start, after having read in the particles, constructed the top-level cells and all their progenitors we do a quick pass to sanitize the whole thing. If a top-level cell has more than 100000 particles (say) but has an `h_max` that prevents tasks associated with this cell from being split we do the following:

 - Compute the geometric mean and standard deviation of the smoothing lengths in that cell.
 - Limit the smoothing lengths of all particles in this cell to mean+4sigma (say).

Recall that the cells are always split so that the scheduler will be able to push down the tasks at a later point. 
This procedure is applied to the top-level cells only.

Does that make sense to you ?

See merge request !258

Cleanup rho dh and use of the EoS functions

A few more rather unimportant cosmetic changes:

 - moved rho_dh to density.rho_dh everywhere.
 - added a force.f term to contain the transformed value of rho_dh. 
 - When computing the sound speed, I am now using the functions defined in equation_of_state.h everywhere.

The particles are the same size in memory as before despite the change. I had to update the tolerances for the tests as rho_dh changed meaning. 

See merge request !255

Fixed bug in HLLC Riemann solver flux calculation.

This fixes #216 (again, but now all cases).

I basically overlooked a tiny detail in the real GIZMO and changed the order of some operations that cannot be interchanged. Square test is now verified to work with all 3 possible Riemann solvers.

See merge request !257