ETL default task changes & separating out study-specific workflows

[mikix]

In anticipation of additional future studies & tasks, it makes sense to re-examine our default set of tasks and services and how you interact with studies. There are two notable changes that just landed:

Dropping covid symptom tasks from the default set

Specifically, the covid_symptom__nlp_results task will no longer be run by default (i.e. if you specify no --task or --task-filter parameter). You can still run it by name or by the covid_symptom or gpu filters.

Going forward, we won't add specific study tasks to the default set. We'll leave that for core FHIR resource tasks.

Renaming the etl-support docker compose profile

Previously, in preparation for running ETL, you would run docker compose --profile etl-support up -d to start the cTAKES and cNLPT services.

But with multiple studies, it doesn't make sense to have a global profile that runs every service that any task could want. One study might need llama2, another might need cTAKES. Instead, we're renaming the etl-support profile to be study-specific: covid-symptom.

It's exactly the same, except you type docker compose --profile covid-symptom up -d (or use etl-support-gpu).

Going forward, as studies are added, we'll make new profiles for the services they specifically need.

tl;dr;

• Covid tasks are no longer run by default
• You no longer need to run docker compose --profile etl-support up -d before ETL runs
• But if you do specifically want to run Covid tasks, you should run docker compose --profile covid-symptom-gpu up -d first