If this reinvents half of postgrest then postgrest is a very small piece of software (I don't think it is). I think the author has restricted themselves to such a small portion of SQL, it's probably not really worth comparing it to that, it's really just a small bespoke language transformer to SQL, and this is not so hard to implement.

But, even if they did reinvent half of postgrest, there are plenty of reasons this might be the right choice given more contextual information, tech decisions aren't made in a vacuum, it just might not be possible to use postgrest.