Fixing contamination_fraction in server mode

[Martin-Rey]

Hi,

This PR is a fix to issue #85 for which the contamination_fraction of haloes made of purely heavy particles would be 0.0 in server-mode.

This is bypassed by implementing a mass_resolution at the simulation level, similar to the spatial resolution already present. A quicker option from the snapshot size is also available if requested.

I also added two units tests on the Changa handler verifying that both the normal and quicker calculations return expected results. I further verified that the behaviour is the one expected from the documentation on three tests zoom simulation, both with and without the quicker flag.

Comments welcome!

Martin

[Martin-Rey]

So the travis builds fails with bizarre errors that I cannot reproduce on my local setup. The source of the errors seems to be

Traceback (most recent call last):
  File "/home/travis/miniconda/envs/2.7_1.13.1/lib/python2.7/site-packages/nose/loader.py", line 418, in loadTestsFromNameaddr.filename, addr.module)

  File "/home/travis/miniconda/envs/2.7_1.13.1/lib/python2.7/site-packages/nose/importer.py", line 47, in importFromPath
    return self.importFromDir(dir_path, fqname)

  File "/home/travis/miniconda/envs/2.7_1.13.1/lib/python2.7/site-packages/nose/importer.py", line 94, in importFromDir
    mod = load_module(part_fqname, fh, filename, desc)

  File "/home/travis/build/pynbody/tangos/tests/test_consistent_trees.py", line 2, in <module>
 import pynbody

  File "/home/travis/miniconda/envs/2.7_1.13.1/lib/python2.7/site-packages/pynbody/__init__.py", line 21, in <module> 
from . import util, filt, array, family, snapshot

  File "/home/travis/miniconda/envs/2.7_1.13.1/lib/python2.7/site-packages/pynbody/util.py", line 29, in <module>
    from ._util import *

  File "__init__.pxd", line 918, in init pynbody._util
ValueError: numpy.ufunc size changed, may indicate binary incompatibility. Expected 216 from C header, got 192 from PyObject 

I suspected it was due to the new version release of CYTHON (Jan 19th) but I can't reproduce this issue at all. Any ideas?

[Martin-Rey]

Update: this building/compilation error from pynbody was also reported here (pynbody/pynbody#503) and seems to originate from incompatible numpy/scipy interactions.

Any preferred directions to fix this?

[Martin-Rey]

Thanks for the comments! They should be addressed in the last commit.