fix(mimic-iv): include phenylephrine infusion variants

[Chessing234]

Phenylephrine infusions recorded under concentration-specific item IDs are omitted from phenylephrine, so affected stays are missing from or understated in vasoactive_agent and norepinephrine_equivalent_dose.

Include infusion IDs 229630, 229631, and 229632 alongside 221749, using the labels reported in #2159. Keep the existing dose conversion and exclude the intubation bolus ID 229789. Regenerate both PostgreSQL and DuckDB outputs.

Fixes #2159.

Validation:

• Added synthetic regression tests for every infusion ID, mcg/min weight normalization, exclusion of intubation/unrelated items, and propagation through the vasoactive-agent/NEE pipeline, including concurrent norepinephrine. These tests run in the existing transpile CI job.
• Both new cases fail on the original source/generated SQL and pass with this change.
• 340 local transpiler, comparison, and regression tests pass; SQLFluff and git diff --check pass.
• Executed the committed PostgreSQL SQL through the same three-concept pipeline on PostgreSQL 17; expected rates and NEE values match.

Validation used synthetic data; I have not independently queried the full credentialed MIMIC-IV dataset to reproduce the cohort counts reported in the issue.