Check that the code and ICs have the same dimensionality

Does what it says on the tin. :)

Implementation of #203 following @bvandenbroucke's request. When reading ICs we check whether the ICs have the same dimensionality as the code was compiled with. We assume d=3 by default to retain compatibility with Gadget-[234].

Branch can be removed.

See merge request !221

Allow the users to change the compression rate of datasets.

Compression is switched off by default for performance.

See merge request !220

Removed calls to fminf and fmaxf in the critical sections of the code.

Is there any reason not to do this ? I have replaced calls to `fminf` and `fmaxf` by ternary operators. That leads to massive speed-ups. 

According to vTune, we were spending large amounts of time inside these functions. 

Is there any reason to use them over what I have done ? Any special behaviour of the functions that we would need in some cases ?

See merge request !219

Disk patch

generates glass-like ICs for a disk-patch (gas in vertical hydrostatic equilibrium with external potential). Runs this disk-patch without assuming gas remains isothermal. See readme.txt

See merge request !217

Allow the users to change the compression rate of datasets. Compression is switched off by default for performance.

Tasks cleanup

Replace all tasks that are not part of the main computation to a `threadpool` which can more simply and efficiently schedule small tasks with no dependencies. See #166.

Peter, can you check that this effectively passes all your tests? Thanks!

See merge request !176

Removing duplicate functions to avoid some merge issues for other branches

Master is broken, so back to original for now.

This reverts commit 409531ab.

Removed now-unnecessary large glass file from the repository. All examples download the required glass files from the web-server.

Fix the minimal version of SPH

Ok, I have now fixed the minimalistic implementation of SPH. That's the one that should be used as a template for other people to implement their stuff. It's overly documented and now works on all the test-cases we have in the examples directory.  

Don't know if you want to look at this or not @pdraper given that it does not affect our default setup at all.

Solves #148.

See merge request !218