Prediction API now returns message specifying if a needed parameter is not
passed to the prediction algorithm, and passes error messages that occur within
the prediction algorithm.
The API passes an error message if a non-existent algorithm is selected, as before. The API now also passes an error message if a necessary parameter to the algorithm is missing, specifying the algorithm chosen and the required parameter. All other errors (e.g., bad DICOM file path) should be caught within the prediction algorithm, and these messages are caught and passed through the API.
## Reference to official issue
Fixes issue #11
## How Has This Been Tested?
These changes are tested through additional tests in test_endpoints.py. To properly test the response to bad DICOM paths, I modified the "predict" functions of the three "trained_model.py" modules to load the passed DICOM path. This loaded image is currently simply discarded after loading.
## Screenshots (if appropriate):
- [x] I have signed the CLA; if other committers are in the commit history, they have signed the CLA as well