Similarity check is too generous for shallow call stacks

[hbrodin]

For two call stacks:

parse_a
harness_func

parse_b
harness_func

The similarity checks in the scheduler will merge them as they are very similar, however for the case I was looking at parse_a and parse_b required individual patches and those POVs should not be considered the same.

[hbrodin]

UPDATE: Actually in my case it was the same type of vulnerability in all of the functions so it kind of made sense in the end. The patcher was able to patch across three different functions to fix it 🥳 However, for a more generic case I think we might have to come up with something more accurate.

[ret2libc]

Definitely, the similarity check is way too simple. TBH also considering all stacktraces in the same function as the same vulnerability is kinda wrong.