I was preparing an answer and I would like to keep it below between the lines. While writing it and in need of an example I came to a point that removes the ground for the question. That is at the bottom at the very point it naturally came into discussion. Nevertheless, the answer in preparation appears of some relevance, too. That's why I post it.
It seems to me that is possible for open (energy & mass exchange) systems in general.
How to compare the entropy of the plant "before and after growing"?
It is difficult to define living things as (sub)systems without including flows or even dividing them further until equilibrium thermodynamics can be applied. For instance, in your example, the growing plant has fixated oxygen and carbon dioxide from the surrounding. So it is not easy to answer while satisfying your last requisite. Or if carbon dioxide and oxygen were already accounted within the system at least the plants itself has changed in itself.
Perhaps a specialist of non-equilibrium thermodynamics can, but I am not. I try to make the discussion more sensible, at least according to my thinking.
I can't recall any non-living chemical system for which entropy decreases during an endothermic reaction whatever coupled to another process. But chemistry flasks are way less organized than organisms.
On the other hand, the comment mentioning whether physics let me think that indeed something similar can happen. A system in meteorology is certainly open, as such it might operate a similar process at least on its temporal scale.
What is sure is that living things apparently do that (processes that are endothermic and reduces entropy) but is definitively clear as well that in parallel they operate processes leading to heat and wastes. This way of seeing does not require the surrounding as you requested - because we could refer to a mature organism that does not grow nor it "does waste away": the correct balance now does occur within itself.
If it suffices for the definition of living things I do not know (manufactured things by living entities not considered).
I think it is a complicated argument that goes normally untreated or leads to debate even among specialists. For instance, there was (is?) a huge debate on photosynthesis, namely if unitary elemental steps of photosynthesis obey the second principle of thermodynamic (not considering efficiency limitations!). By the way,...
.... considering the efficiency of the process remove the ground for the question, at least using the plant as an example !!!
Not all photons contribute to the growth of the plant. Actually, a very tiny fraction of them leads to organized molecules that store energy. Most of the photons just undergo much degradation into heat that warms the plant. So we cannot say that the process reduces entropy.
A non-living analog is a photovoltaic cell. If we look at the outcome of interest it also apparently reduces entropy while absorbing energy. However, also in this case and without considering the surrounding, most of the photons absorbed just eat up the device. Again the low entropy is just apparent if we do not look carefully enough.
Moreover, I noticed the following. Most likely a solar cell is a manufactured object. However, in principle, the right combination of just a few naturally occurring materials might layer in the right way and gives a whatever inefficient photovoltaic effect. This would be a system leading to an apparently endergonic and entropy reducing process when we omit the wasted electrons and as such it independently negatively answer to the question. At least inanimated systems can behave as the living ones, in this specific respect. No (sub)system anyway does what described in the question if carefully inspected.
edit "entropy decreases during an endothermic reaction whatever coupled to another process" was before saying "exothermic" that was a trivial quid pro quo, not the meaning.
