It would be nice if we knew how to actually _write_ the SBML model back to
disk. But that's a completely different story (and package).

support n_genes accessor

Prepare accessors for genetic and chemical information

The Big Rename of LinearModel to CoreModel

make the abstract interface easily implementable

We don't need the `where` clauses if there's just one MetabolicModel

Design decision: Because we can't have typeclasses even as type tags, we can
not enforce that the functions will be able to statically typecheck that the
given model can in fact return e.g. a genetics-relevant information. Failing
deep in the code because of missing implementation is not an option either.
This is the third choice -- if something's missing, we should just return a
correctly typed "empty" state. Arguably this is good also for converting and
writing models, where it reduces the amount of mixed possibilities for the
conversion.

add automated documentation tests

add docker img for deploy doc stage

Small fixes

add codecov and bot for formatting