Prototype automated forestTIME updates (?)

[diazrenata]

See if I can set up a Docker container + gh actions workflow to automatically generate the forestTIME tables and push them to a private Zenodo archive

[diazrenata]

See https://github.com/diazrenata/automatic-trees for a demo repo that:

• downloads CT data from DataMart
• makes tables
• commits and uploads duckdb file to the github repo

This is a proof-of-concept for running the database update workflow on GH actions. To scale up to the whole database, the following will need to change:

• Probably, the file downloading will need to be streamlined. For example, downloading each .csv file, unzipping it, and removing the .zip. Also, possibly creating the database tables in series and then deleting the .csv files. For example, download all the TREE tables, create the TREE table in the database, and then delete _TREE.csv; then* repeat for PLOT, etc.
• The full database is about 6 GB and cannot be committed to a GH repo. Instead we should either upload it to box via a service app, or push it to an online data repository (figshare or Zenodo).
• Finally, there is the possibility of some memory overload issues in the course of running the pipeline on the whole US given resource limits for GH actions VMs