Task sub split

This implements #175 
While at it I also removed some dead code (including #174) and cleaned up the linking of libraries in the Makefile. For some reasons we had still blas in lapack in there.

Finally I have also removed some commented-out old gravity stuff. That should make the diff for the gravity branch much easier to read when it becomes time to merge that in.

See merge request !180

Unified the atomic calls everywhere using the defined macros

Some more code clean-up. I have modified all the calls to __sync_* to use the defined atomic macros. Makes the code easier to read. 

Note that in the case of the timer function, moving to the defined macros changed the return value (++i instead of i++). But we don't use the return value anyway. 

Re-assign to me if you are happy to have this merged in.

See merge request !181

Queue fix

Separate adding and getting a task. Uses a double-ended queue to buffer incoming tasks. The buffer is then added to the actual queue whenever queue_gettask is called.

Peter, can you check if this doesn't break anything? Thanks!

See merge request !178

Style

Here is an improvement to the `runner_shift` poor style. I changed the semantic of the array to better reflect what the array actually contains. As a bonus, the formatting script treats it correctly.
Also it should have been a double and not a float.

While at it I also cleaned out the 'do_gsort()' function which is not used and won't be in the future. 



See merge request !177

Fix initialisation of particles

Hi Peter, 

That should fix it but breaks the MINIMAL_SPH case. I'll re-open the corresponding issue.

Thanks !

See merge request !179

add an incoming waiting list for the queue, makes adding new tasks faster as it does not always require the queue lock.

Optional parameters

This is an implementation of #137. 

I have removed some default parameters from the usual test case YAML files as well. Makes them easier to maintain and simpler to use. 

@pdraper Are there any default settings we should use for domain decomposition ?

See merge request !175

Formatting

"Fix" to #167. Added a script to run clang-format-3.8 with the appropriate style file. Is the 'Google' style with one modification: one can have an empty line at the start of a block. 

All the includes are now in alphabetical order without creating any compilation issues any more.

See merge request !174

Parameter file in snapshot

Sorry @pdraper.... one more for you...

This is an implementation of #140. The parameter file key-value pairs are written to the snapshots. 
Changes are:
 - Write the parameters to the snapshot.
 - Reduce the size of the parameter file structure by changing the maximal amount of lines it can contain.
 - Added an error message for when this limit is reached.
 - Change the type `lock_type` to `swift_lock_type` in `lock.h`

This last change is because `lock_type` is used in some MPI function prototype and our "#define" interferes with some of the includes if you are not careful about ordering. That solves the problem by naming the lock type in a unique way.

Apologies for the rather large changelog that this generates for 5 actual lines of useful code... 

See merge request !173