Add single-call concept-to-concept similarity query

[gkennos]

Motivation

Currently, finding concepts similar to a known concept requires two steps: (1) fetch its embedding from the DB via get_embeddings_by_concept_ids, (2) pass the vector to get_nearest_concepts.

Pitch

This should be a single method.

Alternatives

No response

Additional context

Proposed signature on EmbeddingReaderInterface:

def get_similar_concepts(
    self,
    concept_id: int,
    k: int,
    *,
    concept_filter: EmbeddingConceptFilter | None = None,
) -> tuple[NearestConceptMatch, ...]