Improved abstract/base cubes for YAML data modeling

**Is your feature request related to a problem? Please describe.**

Hey, I've been using Cube for about a month now, so I might be missing something, feel free to correct me!

I have a classic inheritance/extend scenario with multiple entities sharing some properties. I'm using the YAML format and I feel a bit limited in how to implement a solid abstraction cube. Currently I have created a `base_entity` cube with all the shared properties that is then extended in multiple `X_entity` cubes.

This works, technically, but there's a few _minor_ limitations. Looking at the [docs](https://cube.dev/docs/product/data-modeling/reference/cube#extends):

> You can also omit the cube name while defining a cube in JavaScript. This way, Cube doesn’t register this cube globally; instead it returns a reference which you can use while combining cubes.

Something like this would be very useful for YAML as well. For my AI-agent use-case this is important because currently the response from `{base_path}/v1/meta` will also contain the base cube which will clutter the LLM's context with unnecessary data.

Sidenote - the [docs](https://cube.dev/docs/product/apis-integrations/core-data-apis/rest-api/reference#base_pathv1meta) claim that:

> Information about cubes and views with `public: false` will not be returned.

Not sure if there's something wrong with my config (maybe `DEV_MODE` affects this), but this doesn't seem to apply

**Describe the solution you'd like**

Add an `abstract` parameter to cubes, which would work similarly as omitting the `name` parameter for JS cubes. Secondarily, this could allow the omitting of `sql`/`sql_table` param for the abstract cubes

Thank you!