Supplemental base64-encoded file for byte sequences and "extra-dangerous" strings

[timmc]

The README calls out several strings that are not included due to destabilizing effects on the repository:

• EICAR test file: EICAR test file #111
• Invalid UTF-8: JSON cannot represent some naughty strings #20
• Null byte: blns.txt in Pull Requests unexpectedly being changed to binary format w/ no diff #28

It sounds like there's an architectural opportunity here: Create a new list of base64-encoded inputs that contains supplemental strings. This could be a JSON file, like so:

{
  "sequences": [
    {
      "description": "Single null byte",
      "data": "AA=="
    }, {
      "description": "Bare FF byte, invalid to decode as UTF-8",
      "data": "/w=="
    }, {
      "description": "EICAR antivirus test file",
      "data": "WDVPIVAlQEFQWzRcUFpYNTQoUF4pN0NDKTd9JEVJQ0FSLVNUQU5EQVJELUFOVElWSVJVUy1URVNULUZJTEUhJEgrSCo="
    }
  ]
}

This would either be presented as-is alongside the currently built artifacts, or be used in a new packaging method that is capable of representing such data.