Optimize search for text that is not case sensitive

[zufuliu]

Currently, case insensitive search is very slow, besides the brute force algorithm, it will do case folding for the searched text and characters from current document each time a comparison is needed.

Commit 334befc and d308d41 replaced brute force algorithm for case sensitive search with the Boyer-Moore-Horspool-Sunday algorithm, which is very faster in general.
Commit 3455e3b added case sensitivity detection for ASCII text, which makes mark occurrences runs faster, see #236.

As of Unicode 13.0.0, there are only 2847 characters (code points) which has case sensitivity, characters in most other languages (scripts) other than Latin, Greek, Coptic, Cyrillic, Armenian, Georgian, Cherokee are not case sensitive, the "Match case" option make no sense for these characters, enable automatic case sensitive search will benefits people handing such text, especially mark occurrences in large document.

[zufuliu]

Implemented by 0c5d082 and 5b89ad2.