The description of each tests (i.e. the text in the describe, context and it blocks) is a bit inconsistent, we should move towards a more standard structure.

I suggest:

• contract()
• describe(<feature/function>)
• context(<context (!), e.g: once deployed, in refund period, when disabled, etc>)
• it(<verb in present simple, e.g. 'accepts payments', 'allows owner to transfer ownership' // avoid using 'should do x', go for 'does x' instead>)
  • Describe what you're testing, not what you're trying to prove: e.g. if testing a transaction fails from a non-owner account, write 'rejects non owners', instead of 'requires an owner account'.
  • Use the terms 'null address' for address 0, 'zero-valued' for txs that don't transfer ether, and 'null [role]' for a role assigned to the null address (e.g. 'allows setting a null owner')

This will allow us to write test chains that read very naturally, e.g.
PausableToken transfer when paused reverts