support crate authors in checking if docs work in their CI setup.

I'm not sure about the best approach here. this is currently more or less a braindump of my current thoughts
about this.

The use-case is: crates wanting to validate their docs-build in their CI.

Currently we're recommending cargo docs-rs, which duplicates our custom metadata reading, generates a custom cargo command to check if the docs-build works.

current caveats:

• the command is not exactly the same, since the logic is duplicated, it might diverge more
• it's not using the docker build-image, so crates that need c-libraries need to install them themselves in the github workflow, and might diverge from our build-image.
• when your build script does network access, it will pass, while docs.rs fails.
• when your build script writes to the wrong folder, it will pass, while dosc.rs fails.

1) just reduce duplication

create & release a subcrate that just contains the cargo-command-creation and uses our metadata library. cargo-docs-rs can use it

2) do our own cargo docs-rs

build & release our own binary as subcrate, use whatever is needed from the codebase.

3) perfect world

build something that

1. uses rustwide & docker
2. uses
3. and uses the correct command.

best case it's in a way that docs.rs itself is using the same component, and the new cargo docs-rs command would just use the same component.

Now, thinking about this: how far away is this from just running docs_rs_builder build crate ./local_path?

Start by comparing the recommended cargo-docs-rs command with docs_rs_builder build crate ./local_path, then read how rustwide and the Docker build-image are used. Define which approach should be selected and verify that CI checks use the same command, metadata handling, and build restrictions as docs.rs.