Fix EZP-21963: create separate "promise-based" build

[wiseman17]

JIRA issue: https://jira.ez.no/browse/EZP-21963

Done.

• Additionaly, I have created separate page for manual testing of the promise-based version, since it became a non-trival matter to include it into a project, and also as means of "integration testing" for the promise-based version and it's external dependency.
• Updated readme a bit
• I had to reconfigure the unit tests, since the Q dependency is not mentioned in define call now.
  I'm not 100% sure, that window.Q = require("../../node_modules/q/q"); is a nice fix, but it works and looks quite nice to me :) @jakobwesthoff, your opinion is welcome here. May be I've even chosen a wrong way to get rid of "manifested" Q dependency?

[dpobel]

to ease reviews (and reading of the raw markdown file), can you please configure your editor to limit the line length at less than 80 characters

[dpobel]

(this comment also meant that you need to cut those lines, I'm sure you editor has a magic shortcut for that)

[dpobel]

In addition to the inline comments, the promise versions should be mentioned in the installation paragraph of the README

[wiseman17]

Fixes and additions implemented.

[dpobel]

I would put this file in test/manual/jsRestClientBundle (and adapt the README) as this is needed by the bundle only.

[wiseman17]

Fixed

[wiseman17]

@dpobel, is it ok to merge this one?
Taking into account your comment about separate commit, of course.

[jakobwesthoff]

I am not 100% sure, why you needed to take out the Q dependency and replace it with the window.Q hack. Can you elaborate on this?

[wiseman17]

Jakob, I've made it when I was trying to get rid of included Q and achieve "flexibility" you were mentioning here: #25
Partial quote:

The reason for not including Q (even int the promise build) was to give the user of the library the highest possible flexibility regarding choice of the used package management (requirejs, manual download, commonjs, ...).

I can't see a way for developer to include Q into project as a manual download without using global Q variable as it is done now. And hence - "window.Q" approach.
But, as I said, quite possible I've chosen the wrong way, and your advises on more proper approach are very welcome!

[jakobwesthoff]

Ahh I see the reason now. I might have created a wrapper for the dependency, which is AMD compatible, to only use require.js compatible dependency management inside of the CAPI code itself:

define("q", [], function() {
    return window.Q;
});

I think this is a little bit cleaner, than using window.Q directly. Otherwise the changes seem okay to me :)

[wiseman17]

@jakobwesthoff, did I get you right? See my latest commit, please.

[sonarcloud]

SonarCloud Quality Gate failed.

0 Bugs
0 Vulnerabilities
1 Security Hotspot
0 Code Smells

No Coverage information
0.0% Duplication