Refactoring: enable access to a HTTP client without parsing, but with redirects and caching

[pietercolpaert]

Related to #4 and #5

To be used for the JSON-LD document loader

[pietercolpaert]

Refactoring should look like this:

parts only for NodeJS

HTTP+HTTPS+HTTP2 nodejs module ↓ cacheable-request module -- make this configurable ↓ follow-redirects module (and before this step, this HTTP is also the base for the HTTP module of follow-redirect module) ↓ Decompress: https://www.npmjs.com/package/decompress-response ↓ This is the NodeHTTPFetcher (this is also the base for the JSON-LD document loader)

Browser + NodeJS version

NodeJS starts from the NodeHTTPFetcher as documented above. The Browser version starts from the browser fetch module, which already implements all of the above. ↓ Promisify result of a HTTP request ↓ Deserializer (popular RDF serializations, conneg, ...) ↓ Extract metadata such as:

• License
• A label per entity
• Paging (hydra + headers)
• Basic page descriptions such as DCAT, DCT, RDFS, FOAF, VOID, schema.org, ... and in headers
• memento
• Time based graphs
• Geospatial indications with geosparql
• ... and make these available in convenient functions

[pietercolpaert]

Important is that we should always be able to stop a request

[pietercolpaert]

Being tackled in https://github.com/pietercolpaert/ldfetch/tree/fix-6