Use the async a_embed_text in _a_get_n_random_contexts_per_source_file.

[andreasgabrielsson]

Context and Purpose
When running golden generation in async mode using a_generate_goldens_from_docs the context generator is still calling the sync version of DeepEvalBaseEmbeddingModel.embed_text. The sync version of embed_text performs a blocking call to the embedding model and will block the entire thread in which the event loop is being executed. This breaks parallelization and hurts performance significantly.

Depending on how the DeepEvalBaseEmbeddingModel is implemented, it also risks crashing the entire application. For example if the implementation of the sync methods looks like in the screenshot below, it crashes with this error message: RuntimeError: asyncio.run() cannot be called from a running event loop.

This PR fixes the error by calling the async method a_embed_text instead.

[vercel]

Someone is attempting to deploy a commit to the Confident AI Team on Vercel.

A member of the Team first needs to authorize it.

[penguine-ip]

@andreasgabrielsson thank you so much!