Since most operations done with the management are "request-response-style" and they can either succeed or fail, it makes a lot of sense to write tests for them as a whole. This would allow test coverage of 90% of the functionality, including all the SQL stuff, without having to rely on very small tests for SQL (which gets tedious very fast) and very big system tests that require setting up a whole BeeGFS.

To achieve this, the dependencies passed into the handlers (e.g. things in Context) must be mockable. This is mainly the BeeMsg connection pool handle. A mock pool is needed to catch outgoing requests and provide a response for the handler.

Testing the whole handlers instead of separate sql functions will also solve https://github.com/ThinkParQ/beegfs-rs/issues/57.