Function ingest 

pub fn ingest(
    corpus: &Path,
    work: &Path,
    max_bytes: Option<u64>,
) -> Result<Ingest>

Walk corpus, parse everything readable, and bucket it by electrical fingerprint.

max_bytes skips any file larger, counted into skipped: the pairwise compare is quadratic in a bucket and a walk re-parses per hop, so one interconnection scale case in an otherwise moderate corpus would own the whole run. None reads everything.

Errors

Fails when the corpus directory cannot be walked or the work directory cannot be written.