Wingman/exactprint gets confused when a comment precedes a metaprogram

[isovector]

whats_it_deep_of
    :: (a -> a)
    -> [(Int, Either Bool (Maybe [a]))]
    -> [(Int, Either Bool (Maybe [a]))]
whats_it_deep_of f =
  -- The assumption here is necessary to tie-break in favor of the longest
  -- nesting of fmaps.
  [wingman| deep_of fmap, assumption |]

Running the metaprogram produces

whats_it_deep_of
    :: (a -> a)
    -> [(Int, Either Bool (Maybe [a]))]
    -> [(Int, Either Bool (Maybe [a]))]
whats_it_deep_of f = fmap (fmap (fmap (fmap (fmap f))))  -- The assumption here is necessary to tie-break in favor of the longest
  -- nesting of fmaps.
  [wingman| deep_of fmap, assumption |]

Notice that the metaprogram didn't get removed!

[isovector]

Actually, this appears to be due to the [wingman| herald appearing on a newline. The same behavior appears here:

wat 
    :: (a -> a) 
    -> [(Int, Either Bool (Maybe [a]))] 
    -> [(Int, Either Bool (Maybe [a]))]
wat f = 
  [wingman| deep_of fmap, assumption |]