SED-ML: upgrade to version L1V4

[agarny]

With SED-ML L1V4 around the corner, it would be good to get OpenCOR to make use of it through the special L1V4 version of libSEDML.

[agarny]

A bit of an issue here... Indeed, I would like OpenCOR to be able to read both L1V3 and L1V4 SED-ML files (since SED-ML support in OpenCOR started when L1V3 was already available), and (only) write L1V4 SED-ML files.

When it comes to reading L1V3 SED-ML files, the idea is to keep retrieving plot features, etc. from annotation (so that we can still read "old" OpenCOR-generated SED-ML files) while those features would be retrieved natively when reading L1V4 SED-ML files.

Now, the problem is that libSEDML used to create L1V2 documents by default (even when L1V3 was officially out), and I had never noticed it. So, many (all?) of our SED-ML files are actually considered to be L1V2 SED-ML files rather than L1V3 SED-ML files, which makes our testing for the level/version of a SED-ML file somewhat moot.

I guess I could still check for the level/version and if it is L1V4 (or newer) then OpenCOR would retrieve things natively otherwise through annotations. As for writing SED-ML documents, I would enforce the level/version to be L1V4 since the special L1V4 version of libSEDML currently creates L1V3 documents by default (see issue fbergmann/libSEDML#56).

@nickerso and @fbergmann, would you agree that this is the right way to go about this?

[nickerso]

I agree, sounds like the way to go.

[agarny]

This will be done as part of our move to libOpenCOR, so closing.