[BUG] does not bounds check iterators

[grizzlysmit]

the following code segfaults:
using while first <= last { instead of while first < last {

To Reproduce
1.

main: () -> int
= {
    words: std::vector<std::string> = ( "decorated", "hello", "world" );

    first: std::vector<std::string>::iterator = words.begin();
    last : std::vector<std::string>::iterator = words.end();

    while first <= last {
        print_and_decorate(first*);
        first++;
    }
}

print_and_decorate: (thing:_) =
    std::cout << ">> " << thing << "\n";

 $ cppfront  pure2-bounds-safety-iterator.cpp2 -p -s
pure2-bounds-safety-iterator.cpp2... ok (all Cpp2, passes safety checks)

 $ g++ -std=c++20 pure2-bounds-safety-iterator.cpp -o pure2-bounds-safety-iterator
 $ ./pure2-bounds-safety-iterator 
>> decorated
>> hello
>> world
Segmentation fault (core dumped)

3. I wanted cppfront to reject it

4. it it passed and segfaulted on running

[hsutter]

Correct. Currently cppfront:

• bans pointer arithmetic for pointerlike types that it knows about, but right now that's just explicitly declared local pointer variables in Cpp2 function bodies (in the future I'll teach it about STL iterator types and other Pointer-like types -- see Lifetime profile section 2.1 for the Pointer type category which includes iterators)
• adds dynamic bounds checking for indexed accesses (but not iterators) if you enable it using the -subscript-checks switch

The long-term goal is to treat STL iterator arithmetic like pointer arithmetic -- i.e., as unsafe because iterators don't know their bounds (in the standard, modulo that some STL debug implementations do attempt to add bounds/invalidation checks).