1. Do "memory leaks" only exist in individual processes, or can they somehow exist throughout all processes?
A "true" memory leak is where some software acquires memory dynamically, for some purpose, and when the purpose is fulfilled, does not release any, or some, of the acquired memory. Although some software runs as a "process", which often "owns" acquired memory, an OS might run software as other "things" that can also "own" memory or use CPU. I.e. one might be unable to say only memory owned by a "process" could have a memory leak, but unlikely all "processes" would have memory leaks (as memory leaks, normally are also software bugs).
2. If the "show process memory sorted" reveals a process to be exhibiting abundant "Freed" memory after abundant "allocated memory", Does that rule out a memory leak for that process?
No, but it doesn't exclude it either.
". . . same model second switch under similar workload, so I could have a minimum baseline."
Sorry, the doesn't really guarantee second switch is representative of what's happening in first switch.
Unless you're very knowledgeable about OS and device drivers type software, it can be difficult to fully appreciate the "interesting issues" that can arise when dealing with real-time events. I.e. if process A runs even a millisecond before process B, all's good, but the converse has a bug. Such can be a real problem to identify and debug, especially when process A, normally, occurs 99.999% of the time before process B, and further, the bug's impact is very, very tiny (like not releasing a single byte of memory it should).
Oh, and even for a situation like you describe, there might not be a "memory leak" per se caused by a bug. IOS, I believe, does not "garbage collect" its dynamic memory. Without it, it's possible memory gets fragmented, and the excessive memory usage is due to control block memory usage (tracking all the allocated and free memory).
