The change works for me.

@Florian3k , if you do not have more changes, maybe we could remove the draft status, what do you think? @mbovel please could you consider a review? Thanks!

There's one more thing to fix. Changes in this PR uncovered some bugs in rendering of this.type in some places. I was busy in January, but I'll be finishing this soon.

Note for @Florian3k, about my adjustments with this-types:
Truthfully I didn’t change much since the WIP PR, since it mostly looked good to go. The added this types in most cases can be helpful in avoiding ambiguities - notably in the tastcase case class IAmACaseClassWithParam[T](x: Documentation.this.T, id: T) I added, without the rendered Documentation.this.T, we would not know which T was referenced. I adjusted test cases classSignatureTestSource.scala and exports1.scala with that in mind.
They aren’t really helpful when it comes to the supertypes in the extends clause - they will always be classlikes, and they will always have links in the signature, so ambiguities aren’t an issue. For those I ended up just adding skipThisTypeInPrefix parameter to TypesSupport. I unfortunately was not able to make it cleaner by adding it as a default parameter (when I did that I had some surpiring issues with path dependent types), so the asSignature methods are overridden to achieve the same effect.

The new changes work for me.

Those new changes seem to be working fine.
It's a bit unfortunate that we have to render Documentation.this.T instead of T (this effect will be also visible in some places in std lib docs), but I understand that correctness beats prettiness.
I think improving that should be tracked in separate issue.