Add keycloak support in AuthorizationCodeProvider

[sbrunk]

Keycloak has a concept called realm for multi-tenancy.

Realms are part of the uri, so a baseUri looks something like this: https://<domain>/auth/realms/<realm-name>. The issue is that baseUri.withPath ignores this prefix, so I've changed that to baseUri.addPath which should still work fine with other providers.

Keycloak also supports looking up endpoints for login, token, logout etc. through a well-known uri:
https://<domain>/auth/realms/<realm-name>/.well-known/openid-configuration but that would a bit require more work so I've just added a static config like for the other providers.

[sbrunk]

Perhaps my assumption that we can just append to the baseUri was wrong, given the failing test:

https://github.com/ocadotechnology/sttp-oauth2/blob/b834b6f791310d388a9efe138bf6c97839e2e7e7/oauth2/src/test/scala/com/ocadotechnology/sttp/oauth2/AuthorizationCodeSpec.scala#L39-L53

Do other endpoints expect this behaviour?

[majk-p]

Thank you for the contribution, much appreciated! I've enabled the pipeline and it seems the tests have failed. The usage of withPath implicitly assumed that the path would be stripped. I think this is not a big deal and the end users of the library can do it themselves. We would just need the tests updated to reflect the change.

[majk-p]

Do other endpoints expect this behaviour?

I think this is just a legacy assumption from when this library was an internal one. I would say this change is fine, just that we'll have to make a breaking release to let the users know they have to strip the path on their own in case someone relied on that behavior.

[sbrunk]

We would just need the tests updated to reflect the change.

I've just updated the test code to do that. The tests should pass now.

[matwojcik]

LGTM, thanks a lot for the contribution!