Use current SYNTROPHY semantics in UFMP, digestion and pleuromutilin review rationales #1744

Follow-up to #752 and graph reviews #1474, #1626 and #1711, prompted by the precommit correction recorded in #1740 / PR #1743.

The actual InteractionTypeEnum.SYNTROPHY definition explicitly does not assert obligacy. It concerns one partner using another partner's product, with product removal making the upstream metabolism favourable. Unproven obligacy must not be the reason to withhold that type. This does not imply restoring it wherever a paper uses loose syntrophy language.

Acceptance:

• CommunityMech:000029 (UFMP): retain all three HYPOTHESIZED nodes, both model arrows and canonical participant subsets. Type the existing source-proposed lactate-transfer node CROSS_FEEDING with PARTIAL/COMPUTATIONAL evidence from the primary Discussion. Clarify the distinction from product-removal feedback, not an obligacy requirement. No new donor, route or topology.
• CommunityMech:000274 (SynCom-YSJ/J digestion): retain the untyped putative electron-transfer node and both qualified arrows. Explain that the abstract's metagenomic associations do not resolve the product-removal mechanism; no obligacy requirement. Preserve positive methane/comparator results and uncertainty about expression and specific transfer routes.
• CommunityMech:000300 (pleuromutilin): keep the measured degradation endpoint and proposed-products node untyped. A measured outcome or differential metabolomics alone is not the product-removal mechanism; lack of obligacy is not the reason. Correct the discussion's nonexistent engineering_design-field reference to the record description. Do not add antibiotic selection, strain optimization or experimental details.
• Recheck but leave CommunityMech:000175 (cellulose MFC) and 000252 (Crystal Geyser) unchanged: their existing CROSS_FEEDING classifications already distinguish product-removal evidence from transfer or residual-pool observations.
• Preserve earlier ledgers/history artifacts. Append new curation events and record histories only for the three edits, add regressions that verify the old-to-new hash chain, regenerate pages and execute the required validation gates.

Sources: fresh primary main bodies for PMID37324413 and PMID32203118; fresh primary abstracts for PMID17695929, PMID18725730, PMID41825563 and PMID42133144. No figure pixels, supplements or raw data independently re-audited in full. Previously retrieved pleuromutilin DOCX text was reread for Tables S1/S2 and the Table S3 scope note, not a fresh main-text retrieval. An accidental PMID33125214 lookup is explicitly excluded as an unrelated vanadium paper.

No new graph structure or Edison submission. Preserve #1715 research status and #183/#1327/#1366/#1716 non-graph follow-ups. Self-adversarial review is not independent approval. Close this issue only after protected-main merge.

Dedup: 1011 all-state issue titles/bodies searched with exact IDs, cited PMIDs and broader syntrophy/obligacy/taxon aliases. All 13 exact-ID/reference/schema matching bodies read, plus #1235; earlier repairs and #752 are acknowledged above, not silently duplicated. The broad local search included ignored/hidden files (440 matching files), excluding .git/docs/site/references_cache; relevant caches separately inspected. Historical comments/PRs were not exhaustively searched.