More variants for embedding-based similarities

[janheinrichmerker]

The STMC1, STMC2, and REG axioms depend on similarities. It might be promising to have similarity variants that are based on word embeddings fine-tuned on test collections like Robust 04.
It is only required to have some word embeddings that PyMagnitude can handle, the rest is already implemented.

To implement new embedding variants, take a look at the fastText implementation here:

ir_axioms/ir_axioms/modules/similarity.py

Lines 200 to 201 in bcf142e

	class FastTextWikiNewsTermSimilarityMixin(MagnitudeTermSimilarityMixin):
	embeddings_path: Final[str] = "fasttext/medium/wiki-news-300d-1M.magnitude"

Define a new similarity mixin extending MagnitudeTermSimilarityMixin and then implement variants of STMC1, STMC2, and REG like this:

ir_axioms/ir_axioms/axiom/term_similarity.py

Lines 39 to 41 in bcf142e

	@dataclass(frozen=True)
	class STMC1_fastText(_STMC1, FastTextWikiNewsTermSimilarityMixin):
	name = "STMC1-fastText"