Check if snapshot directory exists and add a digit to the output

Fix issue #321 

See merge request !368

This is a requirement of fopen()

Remove 8-cell limit in gravity + non periodic case

Fix issue #323 

See merge request !367

Also use errno to get the detailed error about the failure

Subset sorting

Massive change in the way we treat sub-cell tasks and sorts in general, i.e.:

 * Sorts and drifts now unlock *all* sorts and drifts above them, this ensures they are executed bottom-up even if intermediate tasks are not active,
 * Sub-cell task hierarchies are traversed consistently during density/ghost/force computations,
 * Sub-cell tasks no longer do the drifting and sorting themselves,
 * Sorts are only re-computed once the oldest sorted index goes stale.
 * Ghosts exist at a lower level in the tree.

See merge request !343

This would have revealed the last MPI bug a but quicker

Correct number of links in engine_maketasks()

Correctly count the number of cell->task links required and only create the necessary number given the engine policy.

This fixes a bug reported by @lhausammann when testing a simple hydro + self-gravity box. 

It will also (very marginally) reduce the memory footprint in the purely hydro cases.

See merge request !361

Better trigger

Fixes #288 

See merge request !362

repartitioning: make the trigger correctly with a value of 2, previously the minimum no. of steps was in fact 3

Also match sanity checks to same as documented allowed values for trigger

Correctly count the number of cell->task links required and only create the necessary number given the engine policy.

Not need to clear sort flag on reception fo particles before density. This done in the cell_unskip_tasks() routine.

Analyse script shows updates and sid

The task plotting scripts have been updated as follows:
 - Output the number of active particles
 - Output the time spent in pair/sub-pair tasks split by sid
 - Implemented @d74ksy's suggestion to put the legend outside the plot area itself.

See merge request !360

Applied @d74sky's suggestion to put the legend of the plots outside of the plotting area to avoid overlap

The task plot analysis scripts now also displays the number of particle updates done in that time-step. It also shows the time distribution for pairs sorted by SID.