RF publish -> push

[adswa]

After datalad/datalad#4206, we need to adjust all book examples that use publish to instead show push.
Ultimately, we want to achieve a setup that is more clever than what we have now:

• ATM, all examples that show publish are code-blocks. This prevents that every time the book is build from scratch, a real publish (which would also require force-pushing, and would lead to problems if anyone but me attempts it) needs to be made.
• Where we want to go to is executable code snippets and configuring the alternative Git identity with an insteadOf configuration to push to a local dataset sibling instead of the GitHub or Gin repositories that are specified. This way we can also test what breaks.

The current commit only quickly-and-dirtily replaces all references to publish with push, setting up the configuration and executable code snippets should happen in the next few days.

[adswa]

Intermediate summary: it works, but given rather inconvenient output rendering (#390), its not good to merge atm.

[adswa]

I think this is okay to merge for now. The first good thing: Everything still works, nothing breaks 🎉
Nevertheless, this is just the start of a refactoring, while I removed all notion of datalad publish, I have only replaced two instances of code blocks with actually executable code. The remaining code blocks with are pushing to special remotes (dropbox, and Github with GitLFS), but I would suggest that I tackle them when I replace/update this section with content about the rclone helper

[mih]

LGTM. Thx for digging through this!