Improve ClinicalTrials.gov ingest

[cthoyt]

This PR improves the ClinicalTrials.gov ingest in the following ways:

1. Automates download (note, I looked into using the new API but it doesn't appear to give MeSH IDs consistently). Current count is 467,838 trials.
2. Add additional fields (this is now more easily configurable with a list)
   • StudyType (e.g., observational, interventional)
   • DesignAllocation (e.g., randomized, Non-Randomized) - transformed to boolean indicating randomized or not
   • OverallStatus (e.g., Completed, Active, Recruiting)
   • Phase (e.g., N/A, Phase 1, Phase 2) - transformed to integer
   • WhyStopped, e.g. NCT06048705 says it was stopped because GSK changed its research priorities
   • StartDate (e.g. November 1, 2023, May 1984) - transformed to integer representing year
   • AnticipatedStartDate - (e.g. Actual or Anticipated) - transformed to boolean

[kkaris]

Looks good overall. I suggested two fields that could get property types set, and that we need to check for NaN's.