I think of this as the abstract concepts of meter and the regular point at which beats can be multiplied or subdivided, and concrete rhythms actually played or notated.
The wiki article shows first how beats can be divided or multiplied evenly.
Then in shows how regular beat subdivisions can be classified into four rhythmic unit categories.
Up to that point the descriptions are all abstract. None of those patterns are presented as music actually played. It's just a way to map out a bunch of "nodes" in time where regular beats can be regularly divided. You might think of that "map of nodes in time" like the grid lines that can be drawn over a painting which delineate linear perspective. Nothing in the actual composition - painting or music - needs to coincide exactly with those lines or nodes, but conceptually they are understood to exist and are the basis for our perception of space and rhythm.
In music, when actual, concrete music (I don't mean Musique concrète) is played, typical rhythms are used which can be associated with the various four rhythmic unit types. So, a rhythm like...
...you don't have merely the beat or a rhythmic unit, but rather a complex combination of duration that can be classified by those abstract categories.
First there is rest on the beginning of the pulse and so it's anacrustic, the next three sixteenth notes are of the metric subdivision category, the next dotted eighth note and sixteenth are of the intrametric category, the following dotted quarter note is of the multiple level of subdivision, and the following eighth note does not coincide with an initial strong pulse so it is occurring as a contrametric category. Then that two bar pattern repeats.
When the two bar rhythm is considered it is not simple the realization of simple beat divisions or multiples, nor merely one of the rhythmic unit patterns. That kind of complex assemble of rhythmic elements is a rhythmic gesture.
Regarding all the commentary about how the Wikipedia article quotes from Winold, I don't have the book, but can view portions of it in Google Books snippets. The original book contains the following line...
p.239 The rhythmic gesture, in contrast to the rhythmic unit, is not limited by the underlying metric structure; its beginning, end, and length are subject to varying interpretations based upon consideration of factors contributing to cohesiveness and separation.
That seems to have been worked into the wiki article not as a direct quotation and with significant modification...
A rhythmic gesture [is any durational pattern that,] in contrast to the rhythmic unit, [does not occupy a period of time equivalent to a pulse or pulses on an] underlying metric [level.]
I bracketed the modifications added in the wiki article.
Notice the difference between the original... "its ... length are subject to varying interpretations" and the wiki wording... "any durational pattern that ... does not occupy a period of time equivalent to a pulse or pulses".
The wiki article introduces a specific notion about length measured in incomplete pulses that is not in the original text.
In fact, Winold seems to make the opposite point: the length of a rhythmic gesture is subject to interpretation.
Laster in the original book there is this statement...
p.252 The gestures themselves range in length from a single measure with just one sounding note to five-measure durational patterns which might be analyzed as consisting of two successive rhythmic gestures.
While that is certainly not Winold laying out a definition, it reads to me like Winold considers a single measure to be a possible rhythmic gesture length. Obviously a single measure's duration would be 'a period of time equivalent to a pulse or pulses on an underlying metric level.'
The idea "does not occupy a period of time equivalent to a pulse or pulses" seems to have been grafted onto Winold's original writing, without clear citation, and resulting in a meaning different that what Winold meant.
Below is the most I can assemble from Google Books snippets.