System information

• Cantera version: 2.5.0a4 (and earlier?)
• OS: Ubuntu Xenial (16.04)
• Python/MATLAB version: Python 3.5.2

Expected behavior

SolutionArray.write_csv writes extra columns in correct order regardless of OS / Python version.

Actual behavior

Some columns are swapped for some OS / Python versions.

To Reproduce

In [1]: import cantera as ct
  ... : gas = ct.Solution('h2o2.yaml')
  ... : arr = ct.SolutionArray(gas, 10, extra={'grid': range(10), 'velocity': range(10)})

In [2]: arr._extra
Out[2]: 
{'velocity': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
 'grid': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]}

In [3]: arr.write_csv('test.csv')

In [4]: !head test.csv --lines=3
velocity,grid,T,density,Y_H2,Y_H,Y_O,Y_O2,Y_OH,Y_H2O,Y_HO2,Y_H2O2,Y_AR
0.0,0.0,300.0,0.081893927638,1.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
1.0,1.0,300.0,0.081893927638,1.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0

Thoughts:

• This is obviously linked to Python dictionaries not (necessarily) preserving order
• An option that preserves an OrderedDict would require a tweak of internal storage, where the following should result in the expected behavior:

arr = ct.SolutionArray(gas, 10, extra=OrderedDict([('grid', range(10)), 
                                                   ('velocity', range(10))]))