RFMIA graph: remove procedure arrows and distinguish imposed constraints from biological rescue #1524

Record:KBase_Models_for_Zahmeeth_Original_PLOS.yaml (000037),5 nodes/5 arrows; https://doi.org/10.1101/2023.08.23.554558.

Remove framework adaptation and model construction as procedural nodes; all five arrows describe workflow or co-occurrence, not biological causation. Retain three scoped computational observations, without CROSS_FEEDING/NICHE_PARTITIONING/MUTUALISM claims. The primary Methods/Results use E.coli iML1515 and B.theta iAH991, unlike the abstract's alternative pair names. Use canonical E.coli/B.theta as explicit node participants, not all four taxa. Aerobic isolated B.theta growth and MI are manually zeroed, so do not call the contrast independently predicted partner rescue. Preserve positive model dextrin growth and generally higher MI as model-dependent outputs. Do not infer reciprocal fitness or measured communication.

Full v1 author-lab PDF main text/captions read; selected Methods/table/equation/architecture/chart pages visually inspected. No model rerun, code or external supplement audit. Exact metadata/source discrepancy remains a separate follow-up.

Dedup:827 all-state issue titles/bodies, broader synonyms,38 ignored/hidden local matches and relevant source/history/test contexts; not every comment/PR. Whole records reviewed. Self-adversarial review is not independent approval. Keep repair issues open until protected-main merge; metadata/provenance followups remain open until separately resolved.