WL#2048: Add function for Unicode normalization

In order to safely and efficiently compare Unicode strings, they need to be 
normalized so that "equivalent text (canonical or compatibility) will have 
identical binary representations".

Unicode Standard Annex #15 (http://www.unicode.org/reports/tr15/) describes the 
four Unicode normalization forms.

I would suggest adding a function NORMALIZE(string, form) to normalize Unicode 
strings (UTF-8 or UCS2) to the specified form.

Sybase ASE has a setting to normalize its Unicode data types, but does not expose 
normalization via a function.

In Mimer, Unicode data is automatically transformed to NFC.

SQL standard has normalization function. An excerpt from sql2009-nov:

 ::=

NORMALIZE
   
[   [   ] ] 
  

 ::= NFC | NFD | NFKC | NFKD


We'll also need the NORMALIZED predicate, to check if a string is
already in a given normal form.

 ::=
   

 ::=
IS [ NOT ] [  ] NORMALIZED