Add some const qualifiers in HashTable foreach macros

[Girgias]

I couldn't apply the const modifier to the ZEND_HASH_MAP_REVERSE_FOREACH() macro as it is sometimes ended with the ZEND_HASH_MAP_FOREACH_END_DEL() which deletes entries from the HashTable (thus clearly violating the const qualifier).

I stumbled upon this when I wanted to add a const qualifier in PECL/csv to a HashTable that I don't modify and just traverse. But the fact that ZEND_HASH_MAP_FOREACH_END_DEL() exists makes me unsure how to proceed.

[nielsdos]

I think this patch is wrong, I wanted to write a loop like this:

	ZEND_HASH_FOREACH_VAL(ht, zv) {
		// Do something with zval
	} ZEND_HASH_FOREACH_END_DEL();

But I can't anymore because the hash table is now const. Also the reverse version of this also doesn't work anymore.

[Girgias]

I think this patch is wrong, I wanted to write a loop like this:

	ZEND_HASH_FOREACH_VAL(ht, zv) {
		// Do something with zval
	} ZEND_HASH_FOREACH_END_DEL();

But I can't anymore because the hash table is now const. Also the reverse version of this also doesn't work anymore.

Feel free to revert, I suppose what should be constant is the HashTable pointer, not the HashTable itself.

I do wonder if it makes sense to introduce some other, const version of it for people that want to be able to const mark their extension functions as they know they are not going to modify the HashTable.

[dstogov]

I think this patch is wrong, I wanted to write a loop like this:

	ZEND_HASH_FOREACH_VAL(ht, zv) {
		// Do something with zval
	} ZEND_HASH_FOREACH_END_DEL();

But I can't anymore because the hash table is now const. Also the reverse version of this also doesn't work anymore.

I think, this never worked before.
FOREACH_END_DEL is useful only for REVERSE iteration to delete the "tail" of the hash table.