Update the public ChannelMetadata viewset to support token filter

[taoerman]

Summary

Implements token-based filtering for the ChannelMetadataViewSet (v2 public API). Previously, token-based channel lookup was only available through the v1 endpoint (get_public_channel_lookup). This change brings that functionality to the v2 API with support for both Channel tokens and ChannelVersion tokens.

• The ChannelMetadataViewSet now has support for token filtering
• Returns adequate values for channel tokens
• Returns adequate values for channel version tokens (with version-specific published_size)
• New unit tests added to ensure token filtering works correctly
• New unit tests verify other filters keep working when no token is provided

References

Fixed #5462

Reviewer guidance

Run Unit Tests

[AlexVelezLl]

Hey @taoerman! It seems there has been a misunderstanding here, for this spec:

We'll need to create a get_queryset_from_token method. This method will receive a token and will build the channel queryset, fetching the Channels with the token matching one of its secret tokens, and ChannelVersions whose secret token matches the token. Then, we will need to annotate this queryset to fill the missing fields that the Viewset exposes or filters, or that have a different name in the contentcuration models. The idea is that all transformations that we do with the current queryset can be made within the context of this new queryset.

We should get the data from the contentcuration models, because the channels we will import using tokens are private, not public, this is why we can't retrieve them from the public models like this return models.ChannelMetadata.objects.filter(id__in=channel_ids), we will need to get the data from the private models, and annotate/adapt the fields that are different/missing on the public models. As a result, we should get an annotated queryset that complies with the api that we are exposing in the viewset values field

[taoerman]

Thank you for the clarification! You're right - the original implementation incorrectly queried private models but then filtered the public ChannelMetadata table, which doesn't work for private channels that don't exist in the public database. I've fixed this by updating get_queryset_from_token() to return Channel querysets directly and overriding serialize() to transform the field mappings appropriately (e.g., main_tree → root, language → lang), ensuring private channels can be accessed via tokens without requiring them to exist in the public ChannelMetadata table. All tests are passing!