Release of new rest API with frontend

[kkaris]

This PR finalizes the release of the new INDRA Network Search utilizing FastAPI for the backend and Vue.js for the frontend.

Main differences from previous service:

Structure of Repository

The modules are structured around the following components:

• Rest API: defined in indra_network_search/rest_api.py, uses FastAPI, which among other things allow for asynchronous handling of requests and comprehensive type checking of incoming requests.
• Data Models: The data models in indra_network_search/data_models use Pydantic's BaseModels and handle quick and easy data validation for incoming queries as well as outgoing results.
• Query classes and algorithms: For each algorithm that runs pathfinding or other queries of the graph structure, there is a Query class set up in indra_network_search/query.py to handle the mapping of an incoming rest query to the options of that algorithm and also get relevant data from outside sources, e.g. curations or hashes associated with edges for mesh-context searches that are needed to run the query.
• Result Handlers: Just as there are query classes for each algorithm, there are also result handlers or managers for each algorithm in indra_network_search/result_handler.py that takes care of assembling and sometimes filtering the data and provide a structure that can be sent as JSON to the client running the frontend.
• Query Handler: The query handler in indra_network_search/query_handler.py is a class that takes an incoming rest query and figures out which query classes need to be instantiated and passed to the Python API.
• Python API: defined in indra_network_search/search_api.py. This API creates an interface for rest queries to the different graph algorithms.
• autocomplete: This module contains classes needed for the frontend autocomplete functionality. It utilizes the pytrie package.

Vue.js Frontend

The frontend is built with Vue.js providing a better structure and easier development than raw JavaScript + HTML. One benefit over using Flask + Jinja templates is the complete independence from the service running the FastAPI server, making it possible to have less downtime and quick patching. The frontend is available in indra_network_search/frontend.

[cthoyt]

How do you guys feel about applying black/isort since this is basically a full re-write? would make it easier to keep a consistent style

[cthoyt]

:shipit:

[kkaris]

Just discovered that I messed up something in the latest docs build off of dev/master, basically nothing shows up: https://indra-network-search.readthedocs.io/en/latest/modules/query.html

Compare to the build from package a month ago: https://indra-network-search.readthedocs.io/en/package/modules/query.html

I'll have to figure that out before merging.

[cthoyt]

Error in doc build:

...
 File "/Users/cthoyt/dev/indra_network_search/indra_network_search/data_models/__init__.py", line 112, in <module>
    class FilterOptions(BaseModel):
  File "/Users/cthoyt/dev/indra_network_search/indra_network_search/data_models/__init__.py", line 115, in FilterOptions
    stmt_filter: List[constr(to_lower=True)] = []
  File "pydantic/types.py", line 246, in pydantic.types.constr
TypeError: constr() got an unexpected keyword argument 'to_lower'

[kkaris]

Error in doc build:

...
 File "/Users/cthoyt/dev/indra_network_search/indra_network_search/data_models/__init__.py", line 112, in <module>
    class FilterOptions(BaseModel):
  File "/Users/cthoyt/dev/indra_network_search/indra_network_search/data_models/__init__.py", line 115, in FilterOptions
    stmt_filter: List[constr(to_lower=True)] = []
  File "pydantic/types.py", line 246, in pydantic.types.constr
TypeError: constr() got an unexpected keyword argument 'to_lower'

That seems very odd, the current version of pydantic has that keyword argument: https://pydantic-docs.helpmanual.io/usage/types/#arguments-to-constr

[kkaris]

~Will hold off on merge until tox test, docs build and build branch is merged: #19~ Skipping #19