Akka rejects requests when UTF-8 charset is explicitly specified in Accept header along with `application/json`

[Leammas]

The problem occurred after updating to 10.0.6 from 10.0.5. I've made failing unit test here:
Leammas/akka-http@194378e

According to iana the "charset" parameter should not affect compatibility.

The problem was introduced here but I'm not sure how to fix it without breaking new Accept-handling behavior

[jrudolph]

The behavior you describe is implemented when the charset parameter is offered using the Accept-Charset header. However, RFC 7231, section 5.3.2 explicitly mentions that charset is an allowed parameter to put into the Accept-header as well so arguably putting it there instead of using Accept-Charset should make no difference.

That means, we might need to treat the charset parameter in the Accept header specially. @raboof wdyt?

[jrudolph]

Maybe we could improve the line at 4758a44#diff-95adf484e1280a1b0d76142e2901b916R98, to check a mediarange charset parameter against the charset given in the fixed charset mediatype?

[Leammas]

Is the solution suggested in the PR #1140 acceptable?

[2m]

Fixed in #1140