bugfix: Make sure we are using latest known scalameta version

[tgodzik]

PReviously, we were using the latest published scalameta version, which was caused by me setting first latest semanticdb version (that is not an issue), but then I also added the same scalameta version in order to not have potential mismatches between the two. This caused a breaking change in scalameta internal code to break metals CI. Now, we set the maximum version to be V.scalameta

[kasiaMarek]

I guess that's reasonable. However, my understanding is that the main point of this was to add a proper version to the metals-sbt plugin, where we don't really care if the version is "<= V.scalameta" and this additional constraint is not needed there. But I'm not sure how important it is for us to be as up to date as possible.

[tgodzik]

I guess that's reasonable. However, my understanding is that the main point of this was to add a proper version to the metals-sbt plugin, where we don't really care if the version is "<= V.scalameta" and this additional constraint is not needed there. But I'm not sure how important it is for us to be as up to date as possible.

That shouldn't be necessary and also the newer version could be broken. We wouldn't run any test to verify anything on the newer version.