Utilize one database instance for multiple classification jobs

[valentynbez]

Improvement Description
If the taxonomical classifier is loaded into memory, it should be possible to make it available to a few processes simultaneously. I assume the main culprit is GIL. This can be omitted through the use of ONNX library.

On the other hand, adding additional dependency comes with a cost, so the question is if the plugin utilizes a lot of scikit-learn functions, or if trained models can be delivered to users without it.
The hard way would be rewriting a custom prediction function in C/C++.

Current Behavior
joblib pickles the classifier and then starts multiple processes in parallel, increasing RAM usage x processes. Databases will grow in size, so it is something to take into consideration.

Proposed Behavior
Multiple classification processes running from a single instance are loaded into memory.

Priority
Low. Machines with a lot of resources are available to users.

References

1. https://onnx.ai/sklearn-onnx/auto_tutorial/plot_abegin_convert_pipeline.html