This is the place to discuss the issues we have with the phenotype module. We would like to revise it since currently the phenotype table does not address well post-composing terms.
Assumptions
1. The phenotype table is the place for storing the actual value
2. The phenotype (the descriptor) is the cvterm
3. Post composing cvterms can be done now in the phenotype table (using observable_id and attr_id), but we cannot reuse it
since it is tied with a specific value.