Fix namespaces in SED-ML output

[luciansmith]

The auto-generated SED-ML that we produce to go with the test suite is invalid: it does not include the SBML core (nor, in some cases, the FBC package) namespaces. The script that generates these files needs to be updated to do this correctly.

Note that the namespace will be different depending on which SBML file the SED-ML points to, since many levels and versions exist for each test.

[luciansmith]

Hmm, maybe it would be worth it to re-write the SED-ML generator in Python. It might be more maintainable going forward, with a more diverse group of contributors?