Add model based hybrid search optimizer

[wrigleyDan]

• add model
• add API endpoint that accepts a query, and returns the best neural search weight based on max prediction
• add necessary things to requirements, API, Dockerfile ...

[wrigleyDan]

The API works technically, so you can call the new endpoint in the following way:

curl "http://localhost:9090/get_neuralness?query=iphone%20case"

And in return you receive the best neural search weight the following way:

{
  "best_neuralness": 0.7000000000000001
}

ToDo:
Wire the endpoint up with the actual query: predict neural weight, and also use it in the query.