xsv is great for a quick sanity checks (i.e. number of columns, unique values counts in a given column) but for a more serious tasks/giant files I switch to either polars or duckdb converting CSV/TSV files to parquet or parquet data sets.
By giant I mean 25G gzipped files with >10^9 rows like these VCFs: https://ftp.ncbi.nlm.nih.gov/snp/latest_release/VCF/