TextWidget/MultiLineTextWidth : Support non-ASCII characters...

...kindof. Our eventual aim is to standardise on representing text using UTF8-encoded `std::string` objects on the C++ side of things, and `unicode` objects on the Python side (see GafferHQ#1208). But here we're using UTF8 encoded `str` (bytes) objects on the Python side as well. Rationale :

- Returning unicode from `getText()` now would be too much of a breaking change, potentially affecting a _lot_ of client code. The better time to make this change will be when we transition to Python 3, when `str` objects _are_ unicode, and everything will be breaking left right and centre anyway.
- We've already got de facto use of UTF8 in Python anyway, because that's what we get from things like `os.listdir()` and `glob.glob()`.
- The main downside of using UTF8 `str` is that when multi-byte characters are present, naive indexing will produce incorrect results. This is evident in the API for `TextWidget.getSelection()` and `TextWidget.setSelection()`, which will not match the indexing of `TextWidget.getText()`. But we don't do much character-level processing of text, so in the majority of cases this is an improvement over the status quo, where `getText()` alone was unusable (throwing an exception).

python/GafferUI/MultiLineTextWidget.py

@@ -82,7 +82,7 @@ def __init__( self, text="", editable=True, wrapMode=WrapMode.WordOrCharacter, f
 
 	def getText( self ) :
 
-		return str( self._qtWidget().toPlainText() )
+		return self._qtWidget().toPlainText().encode( "utf-8" )
 
 	def setText( self, text ) :
 

python/GafferUI/TextWidget.py

@@ -65,7 +65,7 @@ def setText( self, text ) :
 
 	def getText( self ) :
 
-		return str( self._qtWidget().text() )
+		return self._qtWidget().text().encode( "utf-8" )
 
 	def setEditable( self, editable ) :