The question posed seeks to establish a line, and people have different beliefs about where this line is drawn. These beliefs are at the very heart of what makes us individuals: it's not a gay scenario unless ... it's a gay scenario if ... it's a gay scenario while ...
Simple logic works well at the level of an individual scenario, that is, simple logic may accurately describe a single instance of data in a very large set. However, this very simple logic rarely works well when applied to larger sets of data.
The problem with such simple logic is that simple logic reduces human sexuality to a few simple rules that work well for developing broad classifications. Yet there are far more sexual situations involving humans than there are rules to describe the behaviour. Human sexuality is chaotic ( in the mathematic sense ... and in general ), and as a result, the data about human sexuality are very general.
Thus, you have a very limited set of descriptions for a wide variety of situations, most of which, like the weather, arise from chaos and can never be described. The search for order amongst the chaos of human sexuality results in very imprecise descriptions and very high error rates.
But the classifications, such as gay, straight, or bisexual, still work very well when applied to large groups. The problem is that you can't say "how gay" someone is with any reliability whatsoever.
I am sometimes amused by the arguments about human sexuality being a 'spectrum' or 'range', when it is in fact, not at all a spectrum. Human sexuality is expressed in the form of events. It is human gender that is a spectrum.