Added the EAGLE_6 test case. This is a very small cosmological volume not representative of the workload of large runs.

Allocate a separate sort array for each dimensions.

As discussed, fixes #334.

See merge request !380

Tiny bug fix in engine_marktask()

In engine_marktask() only check whether to rebuild the space for density pair/sub-pair, not for all sub-types of pair/sub-pair. Leads to problems with mixed gravity/hydro runs otherwise.

See merge request !391

In engine_marktask() only check whether to rebuild the space for density pair/sub-pair, not for all sub-types of pair/sub-pair.

Fix problems with empty hydro tasks.

Fixes #333.

See merge request !379

Fixes to the threadpool

Don't free the `tp->threads` struct when not allocated (1 thread).
Also set the function when logging one thread (otherwise we don't 
get the function names in log).

More testing to check 1 thread behaviour and different numbers of
threads. Tweak output to make progress more obvious.

See merge request !390

Improve test to check a number of thread counts, including 1. Improve output and capture of log files

Only free tp->threads when allocated, set the mapper function for logging when only using one thread

Make sure that a repartition type of none does nothing

Previously when running without METIS support builtin a drift all was
triggered each step as the forcerepart flag was not cleared. That is
now fixed.

Also there is no point in calculating whether to repartition or not
if it will never be done, so avoid that.

See merge request !388

Make sure that a repartition type of none does nothing, including triggering side effects like drift-all each step

and they are not for one thread

and they are not for one thread

Threadpool task plots2

Should fix the problems in the rolled-back merge request of the previous branch, `threadpool_task_plots`.

Which adds the following:

 - --enable-threadpool-debugging to the ./configure flags,
 - -Y to main.c to enable dumping threadpool timers much like the task timers.

This generates threadpool_info-step%d.dat files every -Y steps which contain the mapper function name, the id of the thread that executed it, the number of chunks processed, and the tic and toc timers for the mapper function.

In this merge we also have new scripts to plot and analyse these files and have a new
technique using barriers to start and control the threads. This gives significant
speedups in start up time and time to collective wait (i.e. coming together at task
completion).


See merge request !385

Documentation of the FMM-MM code



See merge request !387

Merge branch 'gravity_documentation' of gitlab.cosma.dur.ac.uk:swift/swiftsim into gravity_documentation

First version of the documentation of the MM scheme. Still need to work out the correct analytical expression of the truncation formula in Fourier space.

Conflicts:
	src/threadpool.c

More conflicts resolved as the merge point seems confused after the
previous revert of threadpool_task_plot branch merge.

Disc patch x

Replaced disc patch potential in z with a disc patch potential in x. Added a 1D version of the disc patch test. Changed the GIZMO time step criterion (fixed a bug in the original criterion, and tried to make the criterion stricter to cope with the disc patch potential). Removed some gravity variables that were no longer used. Put the flux limiter in a separate file and made it configurable.

See merge request !382