2.1 regression introduces invalid compartment and parameter references

[Schmoho]

This occurs on model https://www.ebi.ac.uk/biomodels/BIOMD0000000807

The error is very serious since it changes the mathematical interpretation of the model.

This method

ModelPolisher/lib/src/main/java/de/uni_halle/informatik/biodata/mp/polishing/SpeciesPolisher.java

Lines 84 to 103 in 113b467

	private void ensureCompartmentCodeFromBiggIdReferencesCompartment(Species species, BiGGId biggId) {
	Model model = species.getModel();
	String cId = biggId.getCompartmentCode();
	SBase candidate = model.findUniqueNamedSBase(cId);
	if(candidate != null && !(candidate instanceof Compartment)) {
	// TODO: this is not graceful
	candidate.setId(candidate.getId() + "_non_compartment");
	}
	if (candidate == null && cId != null && !cId.isEmpty()) {
	logger.debug(format(MESSAGES.getString("CREATE_MISSING_COMP"),
	cId, species.getId(), species.getElementName()));
	var compartment = model.createCompartment(cId);
	compartment.setConstant(true); // required attribute
	new CompartmentPolisher(polishingParameters, registry, getObservers()).polish(compartment);
	}

has two errors:

1. It interprets the species ids normalized_cancer_cell_a, normalized_cancer_cell_n and normalized_cancer_cell_g as BiGG IDs and tries to create compartments a, n and g.
   This behaviour was preexisting. However, this is actually unwanted here, as the IDs do not in fact represent BiGG IDs.

i.e.

<species compartment="compartment" id="normalized_cancer_cell_a" [...]>
	 [...]
</species>

becomes

<species compartment="a" id="normalized_cancer_cell_a" [...]>
	 [...]
</species>

2. Since a global parameter with id a exists on the model, the method changes its id to a_non_compartment, but references to the parameter in variable attributes on assignmentRules are not updated. This behaviour is reckless and a regression that I introduced in the 2.1 branch.

i.e.

<parameter constant="false" id="a" metaid="COPASI32" name="a'" value="0">
        [...]
</parameter>

becomes

<parameter constant="false" id="a_non_compartment" metaid="COPASI32" name="a'" value="0">
        [...]
</parameter>

So two fixes are required here:

1. we need to do better in detecting BiGG IDs
2. we cannot change the IDs of non-compartment objects if a new compartment is created from a BiGG ID, but need to adapt the name of the new compartment instead