Issue: https://github.com/php/php-src/issues/11934
Author: SerafimArts
### Description
The following code not working now:
```php
<?php
$uint32 = \FFI::new('uint32_t');
$uint32->cdata = 0xDEAD_BEEF;
// Struct
$ffi = FFI::new('struct { uint32_t field; }');
$ffi->field = \FFI::new('uint32_t');
```
Or more real world usage:
```
<?php
$ffi = FFI::new('struct { uint64_t field; }');
$value = \FFI::new('struct __attribute__((__packed__)) { uint32_t lo; uint32_t hi; }');
$value->lo = 0x42; // First 32 bytes
$value->hi = 0xDEAD_BEEF; // Second 32 bytes
// Because PHP (x86) does not support uint64.
// On 64-bit PHP, this can be done with binary conversions (int64 -> uint64),
// but this is not so convenient.
$ffi->field = \FFI::cast('uint64_t', \FFI::addr($value));
```
Resulted in this output:
```
PHP Warning: Object of class FFI\CData could not be converted to int
```
I think it makes sense to do automatic boxing/unboxing to int, at least in the same way as it is
done with GMP. However, I still have no idea how to implement this without breaking backwards
compatibility =\\\