feat: add HTTP extractor

[sudo-suhas]

• Add HTTP extractor, a generic extractor that uses a script capable of
  emitting any type of asset.
• Fix config error field key to support nested fields.
• Disallow import of os module from scripts.
• Remove unnecessary init method on application extractor.
• Add test helper to assert slice of protos.

Depends on #434, that needs to be merged first.
Closes #413

[StewartJingga]

overall LGTM, but I found a case where user might want to make another call for each item to get more metadata.

Let's say we want to fetch jobs metadata from https://example.com/jobs. That endpoint will give us list of jobs that we will convert into Meteor assets. Unfortunately, to get other metadata of those jobs such as lineage, we need to make another call to https://example.com/jobs/job-id for each job. This will not be possible right now in current state.

Is it possible to have that capability in the http extractor? And yes this is outside of this PR's scope, but asking it here in case there is actually a way to handle that.

@sudo-suhas

[sudo-suhas]

The case you mentioned is definitely valid but we will need to think about it a bit more and implement it carefully. There could be security implications; at this point, it is not entirely clear what the scope of recursive HTTP calls and handling would be. There is no easy way to expose it currently. I would consider it outside the scope of this PR and prefer addressing it later.