Description:
------------
intl package unneededly and incorrectly changes certain characters from greek polytonic unicode subrange to characters from greek monotonic subrange.

Here is code:

$text = normalizer_normalize( $text, Normalizer::FORM_C );

Here is the list of incorrectly changed characters:

1) ά 1F71 changed to ά 03AC

2) Ά 1FBB changed to ά Ά 0386

3) έ 1F73 changed to ά έ 03AD

4) Έ 1FC9 changed to ά Έ 0388

5) ή 1F75 changed to ά ή 03AE

6) Ή 1FCB changed to ά Ή 0389

7) ί 1F77 changed to ά ί 03AF

8) ΐ 1FD3 changed to ά ΐ 0390

9) Ί 1FDB changed to ά Ί 038A

10)  ό 1F79 changed to ά ό 03CC

11) Ό 1FF9 changed to ά Ό 038C

12) ύ 1F7B changed to ά ύ 03CD

13) ΰ 1FE3 changed to ά ΰ 03B0

14) Ύ 1FEB changed to ά Ύ 038E

15) ώ 1F7D changed to ά ώ 03CE

16) Ώ 1FFB changed to ά Ώ 038F

Although characters before and after change most often look the same, they are *different* characters, from different unicode subranges and used differently (in texts of monotonic Greek used chars from the right side of the list and in texts of polytonic Greek are used chars fron the left side). 

One bad result of described change of characters is this: the search for polytonic greek string containing characters in the left side of the list fails after such change.

Test script:
---------------
$text = normalizer_normalize( $text, Normalizer::FORM_C );


Expected result:
----------------
Characters from list above should remain unchanged when this command is executed:

$text = normalizer_normalize( $text, Normalizer::FORM_C );

Actual result:
--------------
Characters from list above changed as described above.