DuckDB for benchmarking

[yaseminbridges]

Implement DuckDB for the benchmarking. This will include some major changes:

• Iterate over a corpus ONCE and store the known gene/disease/variants and refer to this table when benchmarking other runs on this corpus
• Instead of storing run results as dictionaries etc these will be input into a table
• Results are stored as tables instead of outputting various TSV files - but can provide an option to write these to TSV