Figure out how to archive Babel outputs

[gaurav]

We probably want to archive Babel files as generated using the UMLS Level 0 data to avoid some copyright/licensing issues.

I propose to archive Babel outputs in the following files:

• config.yaml (used to generate this Babel run): 7K
• compendia.tar.gz (Compendia files): 20G
• synonyms.tar.gz (Synonym files): 34G
• conflation.tar.gz (Conflation files): 171M
• reports.tar.gz (Reports): 124M
• intermediate.tar.gz (Intermediate files): 8.5G
• parquet.tar.gz (Parquet files from duckdb/parquet): 126G
• kgx.tar.gz (KGZ files): 27G
• metadata.tar.gz (top-level metadata files): 11K

Alternatively, we could combine the core Babel outputs as:

• babel-outputs.tar.gz (Compendia, synonyms and conflation files):

We should not archive the following directories:

• duckdb/duckdbs (DuckDB files): 171G
• duckdb/parquet (Parquet files): 146G
• sapbert-training-data/ (SAPBERT training data): 24G
• logs/ (Logs from errors in previous runs): 2.6G
• .snakemake/ (Logs, config, etc.): 16M