marlin: support tp>1 when group_size==-1

[danieldk]

What does this PR do?

When group_size==-1 we have just one scales group that is shared between
all quantized matrix rows. So, in this case we should not try to get the
scales sharded by row.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[Narsil]

Should we maybe put the integration test with num_shard=2 to check behavior here?

Edit: It's already tested, it's the num_groups=-1 variants that are not checked (it's ok to not test it all the time)