Implement benchmarks that compare co-log with other logging frameworks

[chshersh]

Currently we have benchmarks that compare co-log with raw putStrLn. However, it could be more useful to have benchmarks that compare co-log with other libraries, including:

• co-log-polysemy
• katip
• monad-logger
• fast-logger
• logging-effect
• hslogger
• di
• log

The following questions should be resolved during benchmarks:

• Question 1: How these benchmarks should look like and what should they compare?
• Question 2: What is the structure of co-log-benchmark package will be?