I am going to answer this question in a completely different way. My answer will not technically answer your question but may provide you with another alternative in building your schedule.
First, I prefer to keep all / most of my work packages as fixed duration since in most cases we as PMs have to communicate a firm delivery date. So managing to duration seems more intuitive to me. Second, MSProject will calculate your duration off of work and resource availability assuming a perfect resource elasticity. It will calculate it the exact same way no matter the work and, unfortunately, tasks have varying degrees of elasticity. So allowing MS Project to calculate this for you will very likely give you erroneous indications of duration. Third, your schedule will be perfectly "leveled" just at the time you punch your baseline; it will be perfectly unleveled thereafter. So worrying about ensuring your resources are not over- or under-allocated is kind of a waste.
Instead, estimate using min, max, and most likely values for both work and duration with the assumptions of your talent at specific utilization levels. In other words, in your example above, you know that talent A is 50% and talent B is 50%. So I might arrive at 200, 450, 800 hours; 20, 40, 80 days as my estimates and choose to use 460 hours and 42 days as my planning values.
I would load the schedule with this task as fixed duration, load the resources against it, and then plug manually the 460 hours and 42 days in the work and duration fields. I would repeat this until all my work is loaded and then run an analysis to find resource over- and under-allocations. I would then modify and fine tune as necessary to get the talent as close to their allowable work levels as possible. I don't allow project to auto level my schedule.
The end result is, you have the hours you want, the duration and delivery dates you want, and the resources are reasonably close to their allowed work schedule. Once the project starts and you load actuals, the allocations of your team will go up and down constantly and you will find it is not worth it to chase it. Their work constraints will simply be an input into your estimation of remaining work as you move along the timeline.