Create tests to evaluate the number of false positives and false negatives

[romain-dd]

What is done in this PR:

• Download ~500 legit pypi packages
• Download ~1000 malicious packages
• Add folder scanning capability to guarddog to analyze packages
• Dockerfile to test malicious packages
• Clean up json result to avoid empty category like "result": {"npm-install-script": {}, "steganography": {}}. So we''ll only get the categories if something malicious/suspicious has been found, if nothing found, we'll get "result": {}

Test round 1:

• In a container, I tested with 100 legit packages, it took ~5min (which is too slow), and found 12 false positives
• On my local machine, same result except it took 1min 30s.

Test round 2:

• Test ~500 legit pypi packages, I got the result after 8 min and found 46 false positives (I didn't analyze them), so this give us a ratio of ~9% of false positives