Introduce poetry for dependency management and CLI run

Description

This is quite a big PR. The purpose for it was to make it possible to run our ETL processes via the command line as well as through PyCharm. This now does work in both ways

Changes

Use Conda for dependencies only where necessary (xarray and cfgrib) and to bring in poetry and python
Use Poetry for the majority of python package dependencies (referred to in pyproject.toml)
Consolidate both API and ETL dependencies into one (no longer use conda)
Simplify PR build as a consequence of using same dependencies
Poetry automatically understands that src should be included on the PYTHONPATH and so removed all references to 'src' in imports

Explanation

The issue with running via the command line was because of our project structure. Our source code was placed in the 'src' directory under several packages (database, api and etl). Our application scripts that needed to reference these python modules had to import each other with 'src' prepended to the statement, because the 'src' directory was not included in the PYTHONPATH (and actually this is the whole point of src layout).

The solution was to restructure the repo and bring in poetry as is advised here. This has the added advantage of making dependency management much simpler and more like npm.

It should also be possible to leverage poetry for build/packaging purposes when we need to

File	Coverage	Status
air-quality-backend/scripts/run_forecast_etl.py	0%	🟢
air-quality-backend/scripts/run_in_situ_etl.py	0%	🟢
air-quality-backend/scripts/run_purge_old_data.py	0%	🟢
TOTAL	0%	🟢

ECMWFCode4Earth / vAirify