Refactor config functionality to support configuration through environment variables

[tlater-famedly]

Description

The ldap-sync tool should support multiple sources of configuration, both a configuration file and configuration via environment variables. Currently only the configuration file is supported.

This can either be implemented manually, or a crate like https://docs.rs/config or https://docs.rs/figment could be used. While we use config in other crates, it is semi-unmaintained, so figment might be the better choice.

Requirements

• Configuration should be read from environment variables and the configuration file.
• Options set in environment variables should override settings from the configuration file if both are present
• Environment variables should use a program-specific prefix to avoid potential conflicts
• Environment variable should use double underscores to separate key (F.x. server.bind_address becomes SERVER__BIND_ADDRESS

### Tasks
- [x] Component can be configured according to the requirements
- [x] All requirements have tests

[mzaniolo]

What should happen if the file doesn't exist? Should it just fail or should it try to read every config filed from the env vars?

[tlater-famedly]

What should happen if the file doesn't exist? Should it just fail or should it try to read every config filed from the env vars?

It should try to fill the config with env vars. The practical use case for this is kubernetes, where we want to do configuration with environment variables as much as possible.