Avoid ANSI version Windows string functions

[zufuliu]

These functions are really designed for Windows ANSI code pages, may not works for UTF-8 encoded text.
Example for StrTrimA() failure on my system:

const char *str = "\xe3\x81\x8c\xe7\x97\x9b\xe3\x81\x84\n";
char buf[100];
strcpy(buf, str);
size_t len1 = strlen(buf);
StrTrimA(buf, "\r\n");
size_t len2 = strlen(buf);
assert(len2 < len1);

where the UTF-8 encoded str is Japanese text が痛い from issue #127.

it seem StrTrimA fails when parameter contains invalid byte sequences in current ANSI code page.
This failure will cause sorting lines to add extra new lines to finally result, like sorting folowing lines on my systems (also comes from issue #127):

が痛い
が咲く

There are other places where StrTrimA or other ANSI string function is used, and may contains other bugs.

[zufuliu]

usages for ANSI version Windows string functions can be found with grep -nrE "\w+A\(".
currently the following ANSI functions are used (StrDupA() is code page independent, so excluded from the list):

• StrTrimA()
• StrStrIA()
• StrChrIA()
• IsCharLowerA()
• IsCharUpperA()

[zufuliu]

A similar bug is using CRT string function (like strchr()) to process Chinese, Japanese and Korean DBCS encoding text (the tail byte is in range [0x31, 0xFE]):

>>> 'ァ'.encode('cp932')
b'\x83@'

[zufuliu]

This remaining StrTrimA() and StrStrIA() are either used for ASCII strings, or failure are accepted,