Is your feature request related to a problem? Please describe.
Throughout the package, the documentation and warnings related to max_delay were written only with a daily resolution in mind. Now that we have the timestep feature, there are some inconsistencies that might confuse users. In particular, it is not 100% clear right now that when the timestep is e.g. weekly, then the max_delay is also in weeks. Part of the problem is that in warning messages about max_delay being too short etc., we report the delay in days, not in timesteps (for consistency).

Describe the solution you'd like
It should be very clear from the documentation that the max_delay is in the units of the timestep. Moreover, in warnings about max_delay etc. it might be necessary to mention this fact too.

Alternatives you have considered

• Enforcing max_delay to always be in days - but that would not be user-friendly, because we would also have to enforce that it is a multiple of the timestep.
• Reporting max_delay in warnings etc. always in the number of timesteps. This however requires some more work, as the internal checks currently revert everything to the daily level. Also, do we just say "the maximum observed delay is 3 timesteps"? Or do we want to write "the maximum observed delay is 3 weeks"? And what do we do if e.g. timestep=14 then?

Additional context
Code-wise, things should already be consistent, see changes in #224.