The case of GW Orionis is illustrative of this issue. It might have a planet tearing through the protoplanetary disks of that triple-star-system, which would definitely be a case of a planet without a clear orbit because the motion of the stars keeps throwing stuff into its orbital pathway. This stuff gets messy.

Considered more broadly, a big part of the issue is that the term 'planet' is trying to define two different classifications at the same time. One is based on composition while the other is based on positional dynamics within the star system. Both of these things are important in that they provide useful information and prior to 1990 these two categories overlapped heavily, with Ceres basically as the only exception. Recent discoveries, both in the Sol System and in other star systems make it clear that this is insufficient. We need both a position based classification: things that orbit the star in a clear orbit, things that orbit the star in a non-cleared orbit, things that orbit a non-stellar object, things in orbital resonance with a non-stellar object while orbiting the star, etc; and a composition-based classification: stars, mostly gaseous objects in hydrostatic equilibrium, mostly rocky object in hydrostatic equilibrium, mostly icy objects in hydrostatic equilibrium, mostly rocky undifferentiated objects, mostly icy undifferentiated objects, etc.

Unfortunately, exoplanet data is still very limited and data on smaller objects in other star systems is basically non-existent - we have no idea how common asteroid/kuiper belts might be - so we're still basically running with a sample size of one. Really, until we can send a space probe to another star and get a proper look at least one other planetary system it's always going to be muddled.