[RFE] New corpus schema

[ingvagabund]

• Store artefacts as tarballs. Compressed json files are a lot smaller then uncompressed ones
• Consolidate artefacts generated by the same extractor. E.g. symbols extractor extracts three pieces of data: api, static allocations and contracts. They should be stored under the same tarball.
• Store relevant artefacts per package, not per project where it is sane. E.g. API of some projects can be quite huge. They should be stored by a project package so one does not have to read entire json to get a subset of data contained inside.

Some available artefacts:

• golang-distribution-snapshot
• golang-ipprefix-to-rpm
• golang-project-content-metadata
• golang-project-distribution-exported-api
• golang-project-distribution-packages
• golang-project-exported-api
• golang-project-packages
• golang-project-repository-commit
• golang-project-repository-info

The artefacts can be broken down into the following categories:

• project meta data:
  • golang-project-content-metadata
  • golang-project-repository-commit
  • golang-project-repository-info
• snapshots:
  • golang-distribution-snapshot
  • golang-upstream-snapshot (TBD)
• project data:
  • golang-project-distribution-exported-api
  • golang-project-distribution-packages
  • golang-project-exported-api
  • golang-project-packages
• distribution:
  • golang-ipprefix-to-rpm