ci: :ferris_wheel: use semantic-release (node) [release:patch]

[deronnax]

why this move, originally?

[darvid]

i ask myself the same question 🙃

few reasons at the time and based on local testing that made sense to me:

• semantic-release != python-semantic-release (we already know this but just for posterity for future readers); semantic-release seems to be mature and have an ecosystem of plugins (i.e. the pyproject/pypi one I tried to use), a standardized .releaserc format, etc. it seemed like a more robust way to do releases moving forward, well maintained, and less complicated
• unless i'm missing something in the python-semantic-release docs, it has no analogue, or well, multibranch is a poor analogue to the workflow configuration that semantic-release supports that would be conducive to releasing maintenance versions on a different branch (to my brain seems simpler, but 🤷 )
• mentioned already but plugins - could be valuable to leverage that ecosystem for gitops and more release automation
• but mostly i was sick and tired of python-semantic-release not behaving as i expected, as reflected in the commit log 😓 (this is a me problem, not PSR problem)

welcome any suggestions or improvements, i think this project is just more annoying to deal with semantic-release setups due to the nature of having to build and cache and test so many cross-platform wheels. it would be so much simpler if we could allow whatever tool (semantic-release, python-semantic-release) to run a build command, but we can't, because we need to build many times but release once.