Crucian carp metadata: recover reported strain identities and reconcile source discrepancies #1353

Non-graph followup from the batch24 source review of CommunityMech:000293.

The record taxonomy notes say no species/strain is given. PMID:41280275 section 3.5 explicitly names Cetobacterium somerae S3, Pseudomonas stutzeri J1 and Paraclostridium bifermentans M1. Ground these source-reported isolates conservatively and reconcile the genus-level canonical roster in a separately validated taxonomy patch. Do not invent strain CURIEs or silently alter GTDB groundings; #746 remains relevant to grounding reproducibility.

The top-level description also asserts proven immune/barrier mediation, which the graph-only repair will qualify. Reconcile that prose with treatment evidence and proposed mechanisms.

Source discrepancies requiring explicit resolution before cultivation expansion: section 2.4 reports seven days of antibiotics then two days of donor administration, whereas 3.3 reports three days of antibiotics then administration on days 1 and 3. Section 2.8 gives 10^8 CFU/mL suspensions; 3.5 gives approximately 10^8 CFU/L per strain. Do not silently choose either unit/schedule. Preserve the existing #529-bound 78-hour co-culture observation and distinguish it from isolation conditions.

Acceptance: exact species/strain evidence and validated taxonomic grounding, bounded top-level mechanism prose, documented unresolved source discrepancies, history and regenerated page. These are not closed by a graph-only patch. Complete primary body/captions reviewed, no supplementary files/images. Dedup includes 687 all-state titles/bodies plus ignored/hidden local records; comments not exhaustive.