fixes deploy mac and tag release

[mekarpeles]

Closes #11019

Fixes deploys

Technical

Testing

Screenshot

Stakeholders

[Copilot]

Pull Request Overview

This PR fixes several issues in the deployment script and adds new tag/release workflows.

• Defines OL_REPO globally and updates all tag_deploy calls to use it
• Changes how the latest tag is determined and adds tag and release subcommands
• Adjusts interactive prompts in deploy_wizard for more intuitive defaults

Comments suppressed due to low confidence (3)

scripts/deployment/deploy.sh:611

• The new tag and release subcommands should be documented in the script’s usage/help section so users know they are available.

elif [ "$1" == "tag" ]; then

scripts/deployment/deploy.sh:568

• The new conditional flow in deploy_wizard (skipping deploy and optionally checking sync) isn't covered by automated tests. Consider refactoring for testability or adding tests for all interactive paths.

    read -p "[Now] Run openlibrary deploy & audit now? [Y/n]..." answer

scripts/deployment/deploy.sh:406

• [nitpick] Removing the PID from the background-start message makes it harder to trace processes during debugging. Consider restoring the PID output for better visibility.

        echo "(started in background)"

[Copilot]

The backslash-newline in the echo command splits the variable name and can break the URL. Consolidate the URL and variable reference on a single line or correctly reference $LATEST_TAG_NAME without a line continuation.

Suggested change

	echo "The Open Library weekly deploy is now complete. See changes here: https://github.com/internetarchive/openlibrary/releases/tag/$LATEST_TAG_\
	NAME. Please let us know @here if anything seems broken or delightful!"
	echo "The Open Library weekly deploy is now complete. See changes here: https://github.com/internetarchive/openlibrary/releases/tag/$LATEST_TAG_NAME. Please let us know @here if anything seems broken or delightful!"

[cdrini]

One bug, and one spot where some duplicated code's been added. Merge at your discretion, I can't decide how problematic that bug is -- we might want to add at least a warning or a notice. I'm a bit tight on time today so can't fix these rn. We can unblock deploys by having Jim make the PID fix locally, that should fix the deploy if his version of bash also for some reason errors with the -1 index.

[cdrini]

This introduces a bug; the HEAD will be from the directory running the script, not from the version of open library being deployed. So the commit we'll be labelling as the deploy won't necessarily be what was deployed.

[mekarpeles]

I think the problem with using temporary directories and cleanup is any time the deploy breaks part of the way through, we don't have any chance to recover and the tag that we made is more or less lost anyway.

[cdrini]

This is duplicating the code in wizard.