In your actual programming experience, how did this knowledge of STACK and HEAP actually rescue you in real life? Any story from the trenches? Or is this concept good for filling up programming books and good for theory?
To me it is the difference between being a "developer/programmer" and a "craftsman". Anyone can learn to write code and see how things just "magically happen" for you not knowing why/how. To really be valuable at what you do, I think there is a great importance to find out as much as you can about the Framework you're using. Remember it's not just a language, it's a Framework that you leverage to create the best application to your abilities.
I've analyzed many memory dumps over the years and found it extremely helpful knowing the internals and differences between the two. Most of these have been OutOfMemory conditions and unstable applications. This knowledge is absolutely necessary to use WinDbg when looking at dumps. When investigating a memory dump, knowing how memory is allocated between the kernel/user-mode process and the CLR can at least tell you where to begin your analysis.
For example, let's take an OOM case: The allocated memory you see in the Heap Sizes, Working Set, Private Memory, Shared Memory, Virtual Memory, Committed Memory, Handles, and Threads can be a big indicator of where to start.
There about 8 different heaps that the CLR uses:
Finding what heap has high allocations can tell me if I have memory fragmentation, managed memory leaks, interop/unmanaged leaks, etc.
Knowing that you have 1MB (on x86)/ 4MB (on x64) of stack space allocated for each thread that your app uses reminds me that if I have 100 threads you will have an additional 100MB of virtual memory usage.
I had a client that had Citrix servers crashing with OutOfMemory problems, being unstable, slow responsiveness when their app was running on it in multiple sessions. After looking at the dump (I didn't have access to the server), I saw that there were over 700 threads being used by that instance of the app! Knowing the thread stack allocation, allowed me to correlate the OOMs were caused by the high thread usage.
In short, because of what I do for my "role", it is invaluable knowledge to have. Of course even if you're not debugging memory dumps it never hurts either!