Clearer points in favour of macro mixing

[johnduffell]

This PR updates the scala 3 migration docs to emphasis the benefits of macro mixing rather than indicating that it's experimental.

--

The background is as follows:

We are trying to update our code gradually to scala 3. In one case we have a large multi project build which produces many semi-related jar files for deployment into AWS lambda.

I am trying to update this module by module, but I quickly hit an issue where play-json had scala 2.13 and 3 versions.

Normally I would just force version 3 and use -Ytasty-reader, however in this case because it's a macro library, the modules remaining on scala 2.13 didn't compile any more.

I proposed adding macro mixing to the library, but this was pushed back on the basis that it wasn't necessary and is not promoted by the docs. See the issue below:
playframework/play-json#954

I think this is not really accurate (is it?) and I think this is going to be a blocker for all of our larger repos in our company, and possibly other organisations.

[bishabosha]

I was surprised to see the phrasing "Experimental" in the old doc. The only caveat there is if the macro implementation method was compiled by Scala 3 then you can't use things like quasiquotes or eval, but if the macro method is in a separate 2.13 compiled module then this is no issue