refactor CSearchDlg::SearchFile

* separate out transcoded and raw processing into functions
For raw processing:
* most used encodings should be: ANSI, UTF8 and UTF-16LE
* try ANSI, UTF8 and UTF-16BE only in char
* treat UTF-16LE and Binary as char for text search/replace
* only try UTF-16LE aligned in wchar_t for UTF-16LE
* try UTF-16LE aligned and misaligned in wchar_t for Binary only
* try UTF-16LE in wchar_t for Binary only when char mode has no result
* fix Replace in UTF-16LE
* text search is text mode for both search and replace

know issues:
* Treating a multi-byte char as single byte chars:
    yields part of it may be matched as a standalone char,
    so requires non-ASCII char grouped for repeats to get accurate results.
    Still, Line and Column maybe wrong.
* Some encodings are treated as Binary, real binary file may need 2 rounds
   processing, backup is overwriten in the 2nd round.

Author: lifenjoiner