Managers seem to gravitate towards heroic performance because of our collective, albeit unreliable, attachment we have placed on it with superior outcomes. It looks fantastic, so it must follow that fantastic things will come of it.
The problem is heroic performance does not lead to heroic results all the time. It can be costly, risky, and is rarely to never sustainable, predictable, repeatable, and replaceable. Heroic performance is served best when applied to chaos, the heat of the battle kind of thing, but chaos is not a working environment that we want. It happens, we cope, and we hopefully return to normal conditions as quickly as possible.
What we need, in both operations and projects, from our enabling resources--including human talent--is reliable, repeatable, predictable, sustainable, replaceable, transferable and very average performance that yields outcomes consistent with predetermined specifications of performance.
Based on the limited OP write-up, I am not going to pass judgment that either A, B, or C is operating in this manner. There are a lot of variables that have to be understood to understand the benefits and costs of how each person is operating.
That said, I think the underlying issue here is a psychological one, in terms of the sweet gravitational pull of heroic behavior. What has to change is more about an organizational understanding of what constitutes a mature operation or project and what employee behaviors would be consistent with that maturity. Heroic behaviors of a few signal an immature project construct, chaos, high risk, and low confidence of delivery. I would not attack the heroic behavior per se but would look at all the rest of the capability drivers at play, e.g., tools, processes and procedures, training, skill sets, staff size, etc.