osteosarc 0.10.0

A smaller, more consistent osteosarc, centered on exploring the data and making
test data (#58).

• Test data in one command. osteosarc reads T1_tumor --assay rna-seq --variant ID --padding 100 --to tests/data streams just those reads from each of a sample's
  BAMs, and saves each as a small indexed BAM named for its source and variant.
  data.extract_reads(..., to="tests/data") does the same in Python.
• Command line. osteosarc timeline --around DATE replaces osteosarc on, and
  osteosarc test-data (generate, list, verify, export) replaces fixtures; its
  select, pack and panel actions are gone (use the Python functions). table and
  discover are gone.
• No silent overwrites. Saving a download or reads into a folder never replaces
  a file with different contents; it stops and says so.
• Python.
  • Dataset loses table (use parse), claims, timepoints, annotations,
    vaccine_names, pipeline_names, receipts, open_variants, and the
    generate_bundle and select_fixtures shortcuts (use the functions).
  • The package's top-level names are the core API. Helpers such as
    list_bucket, read_records and SNAPSHOT_SOURCES now come from their modules.
• Names. The SV interest catalogue is now SV candidates: load_sv_candidates()
  and the sv-candidates-v1 panel. FASTQ folder tables name assays and platforms
  as everywhere else, with a library column (gene expression, TCR, BCR, antibody
  tags, long reads).
• Docs.
  • Page addresses match their titles (command-line, python-api, snapshots,
    corrections, testing, map2, sv-candidates, test-data, openvax).
  • The three pages about moving code onto osteosarc are merged into two.
  • Every page has much less code formatting.