Simplify how environment variables and config files are used

[sandertan]

It would also be nice to simplify how the environment variables and config files for layers lower than the docker-compose are used. Currently they are passed on both using envs/env-file and environment: in the compose. There's also some discrepancy in how different compose files do this. We could move the environment variables from envs/env to .env-example and load them in the container with something like:

environment:
  - MEDCAT_CONFIG_FILE=${MEDCAT_CONFIG_FILE}

Related to this:
Can we clarify (and possibly simplify) how the config for MedCAT is used (currently configs/base.txt)? The recent versions of MedCAT contain a config within the CDB file, is this then the default configuration for a MCT project? What happens when the CDB config differs from the config at instance level? It would be nice to document the precedence. Also, we can document which parameters are relevant for which stage in MedCAT, like:

Parameter	Creating CDB/Vocab	NER+L	MCT
CDBMaker.multi_separator	x	-	-
General.make_pretty_labels	-	x	-
General.spell_check_deep	x	x	x