In case the path ending with the activity ID has more trailing path elements, we don't fall
back to the default route but just ignore them (and save them in the router state).

This allows passing in extra information via the path without breaking the routing,
and in case we want to forward the routing to the activities in the future we can use
the extra path elements we store for that.

For this to work with our current logic we have to stop comparing path strings and compare
the resulting computed state everywhere instead.

It has new API which we want to use.

This got added in scoped-element 2.1.0 and is more or less the same
as getShadowRootDocument()

Newest Ubuntu switched Firefox to a snap, which breaks karma.

Try to use playwright to download and cache firefox/chromium
and use those for karma.

Compared to puppeteer where we did this some time ago this caches
the browsers globally, and only downloads them when we start the tests.

In case the scoped registry polyfill is loaded we need to create
elements inside of the shadowroot.

getShadowRootDocument() returns the polyfilled shadowroot, or in case
the polyfill is not loaded just return the global document, which works
in case there are no scoped registries.

In case we get something like https://github.com/open-wc/open-wc/pull/2389
this will be usefull. Also it makes IDE completion work + eslint.