[clang-tools-extra] [llvm] [clang] [analyzer] Trust base to derived casts for dynamic types (PR #69057)

[Balazs Benics via cfe-commits]

steakhal wrote:

To illustrate the case of my previous argument, here are two examples:
https://godbolt.org/z/5vWadfPM9
```c++
// base.h BEGIN:
class Base {
public:
  virtual int fun() const = 0;
};

class Derived1 final : public Base {
public:
  int fun() const override { return 1; }
};
// base.h END

Base *spawn(); // Defined in "secondary.cpp"

template <class T> void clang_analyzer_dump(T) {}

int main() {
    Base *p = spawn();
    int n = p->fun();
    clang_analyzer_dump(n); // conj; and never "1"
    int z = 100 / (n - 2);
    (void)z;
}
```

And here is the example with definition of `spawn` inside a different translation unit, which would lead to a division by zero bug at the definition of `z`.
https://godbolt.org/z/eKMWvTPe6
```c++
// secondary.cpp
#include "base.h"
class Derived2 final : public Base {
public:
  int fun() const override { return 2; }
};

Base *spawn() {
    return new Derived2();
}
```


https://github.com/llvm/llvm-project/pull/69057