Ocrd

> * version bump
> * added 3.10,3.11 classifiers

[workflow-configuration](https://github.com/bertsky/workflow-configuration) [49c743d](https://github.com/bertsky/workflow-configuration/commits/49c743d)..[cb923f7](https://github.com/bertsky/workflow-configuration/commits/cb923f7)

> * add mets-copy-agents.xsl
> * ocrd-import: avoid removing non-empty directories on rollback (conversion results / preexisting images)


[v2023-02-06](https://github.com/OCR-D/ocrd_all/releases/v2023-02-06)

Fixed:

* No need to set `TESSDATA_PREFIX` anymore, 353

[core](https://github.com/OCR-D/core) [6331433](https://github.com/OCR-D/core/commits/6331433)..[ee92cfc](https://github.com/OCR-D/core/commits/ee92cfc)

> * docs: fix simple typo, instalation -> installation


[v2022-06-03](https://github.com/OCR-D/ocrd_all/releases/v2022-06-03)

Fixed:

* Work around issues with git's new `safe.directory` setting interfering with `sudo git`, 321, 323, 324

Changed:

* Disable `cor-asv-ann` and `ocrd_keraslm` which require old tensorflow-gpu unless it is available, 325
* Dockerfile: use `$VIRTUAL_ENV` instead of hard-coded paths or `$PREFIX`

Added:

* Start building on Ubuntu 22.04 in addition to 20.04 and 18.04, 318

[core](https://github.com/OCR-D/core) [ecdb840](https://github.com/OCR-D/core/commits/ecdb840)..[60697e4](https://github.com/OCR-D/core/commits/60697e4)