Approximations when resolving qualified names

I have the following code:

import Universum

import Control.Monad.Except (MonadError (..))

foo = runExceptT

And I have this .hlint.yaml

- warn: { name: "Use 'runExceptT' from Universum"
        , lhs: Control.Monad.Except.runExceptT, rhs: Universum.runExceptT }

When I run hlint I get:

$ hlint example.hs 
example.hs:5:7: Warning: Use 'runExceptT' from Universum
Found:
  runExceptT
Why not:
  runExceptT

1 hint

But the thing is that runExceptT is exported from Universum here, because MonadError class doesn't even have runExceptT method, but hlint (probably because of wildcards) assumes that runExceptT is used from Control.Monad.Except module.
If I remove these (..) from import then `hlint doesn't throws any warnings.

I'm on 2.0.11 version of hlint.

Reproduce the report with the shown example.hs and .hlint.yaml using HLint 2.0.11, comparing imports with and without the (..) wildcard. Trace how the qualified names in the custom hint are resolved, then verify that HLint no longer emits the misleading warning while still recognizing a genuinely qualified-name match.