:bar_chart: Speed up GHE & GBD

[Marigold]

What started as performance optimizations of GHE ended up as multiple performance optimizations across ETL.

• pr.concat and pr.merge gracefully handle categoricals and don't convert them to objects
• Add owid.datautils as a dependency to owid.catalog
• Silence annoying DepreciationWarnings when upserting to MySQL
• More efficient ds.save(), especially for large datasets
• Optional repack when saving dataset
• Let etl d profile handle nested functions

This speeds up GHE & GBD by about 40% (mostly because of categorical variables). I was thinking whether this could cause issues for other datasets and whether I should increment ETL_EPOCH to test them all, but I since it only affects categorical variables, I think it should be fine.

[owidbot]

Quick links (staging server): Site	Admin	Wizard

Login: ssh owid@staging-site-ghe-performance

chart-diff: ✅

No charts for review.

data-diff: ✅ No differences found

```diff Legend: +New ~Modified -Removed =Identical Details Hint: Run this locally with etl diff REMOTE data/ --include yourdataset --verbose --snippet ``` Automatically updated datasets matching _weekly_wildfires|excess_mortality|covid|fluid|flunet|country_profile|garden/ihme_gbd/2019/gbd_risk_ are not included

Edited: 2024-06-25 09:14:56 UTC Execution time: 13.73 seconds