Bugfix: path and fx handler

[samroberton]

If there is no :db effect produced by an effectful handler (presumably
because the handler is concerned with some other form of side-effect
than a DB update), then the path interceptor should not attempt to
update the DB value it returns.

[mike-thompson-day8]

@samroberton a test was failing (false looked like a nil) so I created this alternative fix 6a172dd
Does it cover the problem usecases you identified?

[samroberton]

@mike-thompson-day8 your commit is definitely a better fix (and yes, it does fix the issue I observed in our codebase) -- especially in that it also allows you to return nil to indicate that you want this whole path to be set to nil.

One suggestion, though: with this facility now available, the debug interceptor should use the new arity as well (on std_interceptors.clj lines 38 & 39). One of the reasons this bug took me a while to track down was that the debug interceptor was telling me that there was no DB change happening, then suddenly bits of my DB were ending up nil -- ie inconsistency between what debug reports is happening, and what actually happens.

[samroberton]

Closing since 6a172dd covers the original bug, and 486c7e6 covers the suggestion to get the debug interceptor to behave the same way.

Thanks for this, Mike.