Issue: Text broken when receive string value with unicode chars.

Reason: There's bug in BsonReader.GetString(int length) method. The method uses a 128-bytes buffer to receive string value. Each 128 bytes would be converted to utf-8 string seperately, but it may be only half of an utf-8 char at the end. So the text would be broken when contains unicode chars.

Hot to fix: Put the complete data received in an StreamMemory before converting to utf-8 string.