Account for merged taxonomy nodes throughout catalog build

Currently, NCBI taxonomy IDs that have been merged into other IDs are only mapped to their new values when determining taxonomy info for each entity type during the Python part of the build (#1434). In order properly allow for outdated taxonomy IDs to be present in catalog input, we would need to account for them throughout the catalog build process, namely:

- Anywhere that entities are joined with NCBI data (notably, organism synonyms, as of #1697)
- Anywhere that entities are joined with each other (notably, in the part of the build currently done in TypeScript)