This question is almost certainly going to be closed "off-topic", but since I have some experience with communities using CWL in the past, I'll provide my bit.
First, a correction to your premise: CWL is a standard, which makes it different to most other workflow tools. It has several implementations. One of the main benefits is that it's platform-agnostic, making workflows highly-portable.
As you can see from their website, the user community of this standard is heavily biased towards research communities. Implementations of the standards are de-facto default toolchains in several smaller communities, mostly in the bioinformatics domain, where workflows can become quite complex.
There is a strong incentive to publish workflows so that
- workflows can be attributed correclty
- reproducibility can be tested.
If you want to use Apache Airflow, there's an implementation of CWL for that