Update Kitodo.Production and scripts for webhook manager version

[markusweigelt]

• Using current master state of Kitodo.Production
• Adjust sql examples
• Improve process dir script to use
• Makefile improvements: building only specific services using SERVICES env, confirm make down

[bertsky]

perhaps this PR should also link to slub/ocrd_manager#66

[markusweigelt]

@bertsky Adjusted to the changed asynchronous behavior of OCR-D manager. Synchronous is now the default.

[bertsky]

LGTM.

Just needs to be updated to new Manager (after slub/ocrd_manager#66 itself is finished and merged), and one minor change may be required (see below).

[bertsky]

Wait! One bug did slip through:

ocrd_kitodo/.env

Lines 22 to 23 in b8a3885

	MONITOR_DB_ROOT_USER=root # root user of OCR-D database
	MONITOR_DB_ROOT_USER=root_password # root password of OCR-D database

Notice the wrong variable name for the pw?

Live this causes:

MongoServerError: Authentication failed.

[bertsky]

Also fixed in c59b4cb.

[bertsky]

@markusweigelt, short explanation: in trying to make this work, I noticed I do need to update the Controller (because our timeout was partially caused by glitches there, not SSH itself) and added waiting for ocrd-database (because that is what surfaced on the CI as well). So I basically had to put all on one card. Hope all this is sufficient now 🤞

[bertsky]

Ok, it finally (!!) seems to work again. For the moment, only locally – but on the CI all we are waiting for is a release of ocrd/core v2.62.3 to PyPI, so ocrd/all:maximum-cuda can be rebuilt (it does already work with ocrd/all:maximum).

I already merged the new Controller, so IMO the next steps should be:

• update the Controller submodule once more (to the merge commit)
• merge the Manager
• update the Manager once more (to the merge commit)
• merge the Monitor
• update the Monitor submodule once more (to the merge commit)
• merge here
• merge OCR-D Manager base version #100 into main
• convert everything to stable