Fahmy T4 graph: remove unsupported mutualism and distinguish predicted enzymes from removal #1368

Source review of kb/communities/Cyantraniliprole_Fahmy_Consortium_T4.yaml (CommunityMech:000369), including the complete primary main body, tables and captions: Fahmy et al., PMID:42354927.

Both existing nodes carry MUTUALISM, but the source provides community outcomes and post-assay enzyme/pathway predictions, not reciprocal member-specific fitness benefits. Table 8's enzyme assignments are not measurements of expression, flux or enzyme necessity. The HPLC result is parent-compound removal in T4-augmented water containing a native microbiota, not demonstrated mineralization, detoxification or six-member indispensability.

Remove unsupported relationship signs while retaining the two nodes, six canonical strains, positive HPLC result and the existing zero-arrow topology. Keep the 98.27% calculation relative to the 57.79 mg/L zero-day residue, not the nominal 60 mg/L spike. Bound computational versus experimental evidence and maintain an exact-node knowledge gap. Preserve the earlier fixes tracked by #869 and #870; this is not a duplicate of those quotation/denominator issues. No new graph structure is needed for this repair.

Deduplication: 700 all-state issue titles/bodies, relevant prior issues, and ignored/hidden-inclusive local review/history searches. Not every issue comment was searched.