Automate testing against LLM providers & models

[planetf1]

We need to look at automating tests against multiple LLM providers & models, and clearly document the coverage, otherwise regressions can be intoduced as the code evolves. (https://github.com/i-am-bee/beeai-framework/issues/746 may be one recent example)

There may be additional providers/models that are not covered, and this can also be documented.