Should we document fields `null`-ability?

[thejcannon]

👋

#/definitions/waitStep/wait having {"type": "null"} is a bit of an outlier (only one other instance, and its #/definitions/dependsOn).

I'm guessing it is documented because wait step docs explicitly call out null as a value you can use. However, from some light testing, BK is totally fine with other fields being null too!

steps:
  - trigger: foo
     depends_on: ~
     key: ~
  - command: bar
     env: ~
     retry: ~

So, curious if, for correctness sake, most fields should be exploded into support null, or if for conciseness they shouldn't.

[thejcannon]

Another example of null in the schema id #/definitions/groupStep/properties/group/type