Incorrect Asan errors handling (frida, windows)

[mkravchik]

Currently, if Frida Asan detects an error, the handling of is not working correctly.
Without continue-on-error, it panics after reporting the error. However, as this is done from inside the hooks, different parts (such as allocator) are under mutex and then the processing gets suck as we try to gracefully clean the allocatoed memory.
With continue-on-error and if iterations are used, when we get to the end of the executor's run_target, we abort. However, Fuzzer-respawner refuses to start a new instance

I haven't touched the project for a while, so I'm not sure if these changes are relatively new.
Unfortunately, I'm not familiar enough with the respawner logic. If we can make it work this would be a working workaround.