Split lambdas and API calls

[danmermel]

At the moment we have one lambda function calling all the different cryptic analysers... this is getting cumbersome and unworkable.

We want to:

1. Create 3 lambda functions, one for each clue type
2. Create 3 API endpoints under the same api gateway, one for each of the above
3. Modify the front end to call the 3 api calls in parallel and combine the output.
4. Look at the Travis deployment issue - The deploy bit we think only allows us to deploy one lambda function, so we may have to write it ourselves in shell scripts