It simply isn't workable to reason from lighter symptoms to "partial immunity." Let's pretend the vaccine didn't exist. Take someone with a mild case of covid. Would you say this person is partially immune? Or consider someone with no symptoms at all--are they completely immune? Since the vast majority of cases are mild or asymptomatic, are you prepared to argue that "immunity" is actually very widespread?
Again, your "some people win the race, some lose" analogy is on the mark. It's pointless to talk about "partial victory"
You've basically answered your own question. At the population level, immunity is probabilistic. At the individual level, it's ~deterministic. Those who are challenged but stay healthy are immune. Those who are challenged but get sick are not immune. If we were to repeat the experiment, the individuals who stay healthy/get sick would largely remain the same. In this ex-post view of things, there's no "in-between." You got sick or you didn't, you're immune or you're not.
From an ex-ante perspective, we would instead try to predict whether a person will get sick or not. The answer to this largely depends on whether the body can generate a response to previously recognized antigens. This again is binary. You have memory B-cells due to primary response or you don't; you're immune or you're not.
Is that true? (I'm not an immunologist; I genuinely have no idea.)
I'd have thought that it was all a statistics game: how many antibodies you have and how fast your white cells can respond to the threat before the virus got a foothold. In that model there is partial immunity: some people would win that race, and some wouldn't.
If I'm wrong about that I'd appreciate a better model. Thanks.