Port improved comp flattening from eikept/jsbml and run tests

[dyrpsf]

This PR addresses part of #251 by porting the improved comp flattening implementation
from the eikept/jsbml fork and wiring it into the current jsbml-comp module.

Main changes

• Replaced CompFlatteningConverter with the version from the eikept/jsbml fork,
  which provides a more complete flattening implementation for the comp package
  (including proper handling of ReplacedElement/Deletion chains, conversion factors,
  and nested submodels).
• Added the helper classes used by the new converter:
  • CompFlatteningIDExpander
  • CompFlatteningIDExpanderTS
• Adjusted the mergeModels method to merge the “current” (parent) model before
  the previously-flattened submodels. This preserves the original parameter ordering,
  so that flattened output matches the existing testFlattening/testN_flat.xml
  expectations (e.g. for test6).

Testing

From this branch I ran:

mvn -pl :jsbml-comp -Dtest=CompFlattenTest#testSpecificFile test
mvn -pl :jsbml-comp -Dtest=CompFlattenTest test

Results:

• testSpecificFile and the individual flattening/internalisation tests now pass for the local testFlattening and testGathering data.

• testAllData now exercises the full testFlattening suite; there is still one mismatch reported via Success Testing Model (this was previously flagged in the test with a TODO comment).

• testInternaliseExternalModelDefinitions_online still errors due to the remote URL (https://sbmlteam.github.io/jsbml/test/data/comp/simple_online_url.xml) returning a 301 HTML redirect instead of SBML; this looks like a test-data/ hosting issue rather than an algorithm problem.

I’ve kept those two behaviours unchanged but can help adjust the tests if you
prefer (e.g. updating the URL to a raw SBML resource or marking the online test
as integration-only).

[dyrpsf]

Hi, this PR was automatically closed because I had to reset and clean up my branches while reverting recent changes.
I’ll reopen a fresh PR once the work is re-applied cleanly.
Thanks!

[dyrpsf]

Hi everyone,

Just a quick update: everything is now set and all issues have been resolved. The changes have been reapplied cleanly, and this PR, as well as all other PRs I previously mentioned with problems, are now in good shape. Feel free to review at your convenience.

Thanks for your patience!

[draeger]

Very good work. My remarks focus on declarations of some data types that could be more generic in a few places.

[dyrpsf]

Hi @draeger ,

I have refactored CompFlatteningConverter.java to use generic interfaces (Map, List) for field declarations and method parameters as requested.

I noticed that OverdeterminationValidator.java and its corresponding test also use HashMap and ArrayList. Would you like me to apply the same type-level refactoring to those files as well, or should we keep this PR focused strictly on the flattening converter?

The project compiles successfully, though some existing network-related timeouts persist in TestExternalModelDefinition.

Thanks!