Fix two stale SEDML export test expectations

[jcschaff]

Two of the five failures in the nightly SEDML_SBML_IT regression are test
bookkeeping, not product defects. Both models are in the slow set, so they run
only on the nightly and never in the merge gate.

biomodel_188880263.vcml — undeclared unsupported applications

The test declared no unsupported applications; the exporter reports eight. Every
application in that model is rule-based, and SBML export does not support
reaction rules, so the exporter is right and the declaration was simply never
filled in. The assertion has been printing the exact lines to add:

declared: []
found:    [Stochastic mRNA Number Sweep, Stochastic Make mRNA Sweep Const Prot,
           Stochastic mRNA Number Sweep Const Nucl Chaparones, Stochastic Make mRNA Sweep,
           Stochastic, Continuous_Low L1DNA, Network Free, Continuous_Med L1DNA]
           …each: "has reaction rules, SBML Export is not supported"

Added all eight, in the file-ordered position the list already uses.

biomodel_34826524.vcml — a known fault that no longer happens

It had SEDML_FAULT.MATH_OVERRIDE_NOT_EQUIVALENT registered, so it failed for
passing:

file SBML.biomodel_34826524.vcml passed SEDML Round trip, but knownSEDMLFault was set
  ==> expected: <null>

Removed the entry.

Verification

Run individually — these need -Dtest.include.slow=true, since both are
slow-gated and are otherwise skipped:

model	before	after
biomodel_188880263.vcml	AssertionFailedError	pass, 51s
biomodel_34826524.vcml	AssertionFailedError	pass, 346s

Note this repo's PR CI cannot exercise either: SEDML_SBML_IT runs only in
regression.yml, and these two are additionally behind test.include.slow. The
checks on this PR say nothing about the change; the runs above are the evidence.

Incidentally biomodel_34826524 took 346s locally against the 129s recorded next
to it in slowModels, so that annotation is optimistic.

Not included

The other three nightly failures are genuine export defects — exported SED-ML
referencing species and parameters by XPath that are absent from the exported
SBML, plus one duplicate SED-ML id. Those deserve a real fix rather than a
knownSEDMLFault entry, which would make a live bug permanently invisible. See
also #1827, which makes those failures legible in CI output (they currently
report as a bare » OmexValidation).

🤖 Generated with Claude Code

https://claude.ai/code/session_01SvqmME7MkRUNEYje5HpiLt