The UTF-8 language environment is apparently broken; internal representation of
wide characters of various origins is probably at fault.
The Unicode characters entered manually in the search prompt are treated
differently to the same characters loaded in the editor buffer from a file.
Search fails to match between the two kinds.

This bug may be related to bug #4063.
Steps to Reproduce:
1. Launch Emacs.
2. M-x set-language-environment UTF-8
3. Open a file containing non-Latin characters in UTF-8 (e.g. a Russian text).
4. Copy a string containing non-Latin characters from the file into the PC
clipboard.
5. Set the caret at the beginning of the file.
6. Activate the Search or Search-and-Replace (M-%) prompt.
7. Paste the string from the PC clipboard into the search prompt, .
8. Run the search.
9. Set the caret at the beginning of the file again.
10. Activate the search prompt again.
11. Enter the same search string manually, or recall the last search string with
the arrow up key and replace one or more non-Latin characters to themselves
typed manually.
12. Run the search.
Actual Results:  
The search in step 8 succeeds, the search in step 12 fails.

Expected Results:  
Both searches should have found the same string.