kasan: optimize poisoning in kmalloc and krealloc

Since kasan_kmalloc() always follows kasan_slab_alloc(), there's no need
to reunpoison the object data, only to poison the redzone.

This requires changing kasan annotation for early SLUB cache to
kasan_slab_alloc(). Otherwise kasan_kmalloc() doesn't untag the object.
This doesn't do any functional changes, as kmem_cache_node->object_size
is equal to sizeof(struct kmem_cache_node).

Similarly for kasan_krealloc(), as it's called after ksize(), which
already unpoisoned the object, there's no need to do it again.

Signed-off-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Link: https://linux-review.googlesource.com/id/I4083d3b55605f70fef79bca9b90843c4390296f2