Introduce a local IntoHeaderName trait to avoid exposing hyper::header::HeaderName

[lipanski]

Related to #179 and the alternative to #180

[andrewtoth]

Hmm but this will panic internally if an invalid header name string is provided.

[lipanski]

Hmm but this will panic internally if an invalid header name string is provided.

yes and that’s consistent with other methods (like setting the status code to something invalid).

what are you trying to achieve actually or what was the intention behind your initial change?

[andrewtoth]

Fair enough. It is more ergonomic this way and like you said backwards compatible, so no longer a breaking change.

what are you trying to achieve actually or what was the intention behind your initial change?

Initially motivated by this #117. I also ran into the issue of not matching multiple header values individually. By first grouping matchers by header name I think it will be an easier lift to match each header value to a specific header matcher for a given header name. Not sure what your thoughts are on that issue, but maybe we can continue the conversation there?