Description of the issue:

Continuation of #102: Currently both matlab and python users can load the model without loosing any fields, but only matlab users can contribute. This issue is to keep track of the progress towards allowing contributions from cobrapy.

Expected feature/value/output:

The model can go through a full I/O cycle with both cobratoolbox and cobrapy without loosing any fields, and keeping the same order in:

• The .xml file
• The .yml file
• The .txt file

Current feature/value/output:

Keeping an .xml constant is possible when using the following two forks (not tested in yeast-GEM):
https://github.com/BenjaSanchez/cobrapy/tree/feat/matlab-compatibility
https://github.com/BenjaSanchez/cobratoolbox/tree/feat/cobrapy-compatibility

However, forcing contributors to install those forks is not ideal. So instead the packages should be adapted to be compatible with one another. In the following there's a non-exhaustive list of said efforts (will be updated sporadically):

• fix: SBO terms for infinity bounds opencobra/cobrapy#949
• fix: inconsistency in the gene name storage in SBML files opencobra/cobrapy#950
• feat: gene SBO terms in XML output opencobra/cobratoolbox#1590
• refactor/sort-sbml (PR pending to cobrapy)
• fix/secure-annotation-links (PR pending to cobratoolbox)
• feat/bound-sbo-terms (PR pending to cobratoolbox)
• feat/cobrapy-flag (PR pending to cobratoolbox)

I hereby confirm that I have:

• Tested my code with all requirements for running the model
• Checked that a similar issue does not exist already