FMSLogo Code

Fix an off-by-one bug in the un-vbarring logic.
This is a regression introduced by recent refactoring.

Fix a bug in SAVETEST where two procedures were defined with the same name.

Fix a typo in a comment.

Update some comments for ltime() for Unicode FMSLogo
The comments referred to ctime(), but this call was replaced with calls to wcsftime() and localtime().

Fix the READCHAR test to work with Unicode FMSLogo

A file with no BOM is interpreted as ANSI, which maps the bytes 128-255 to values other than 128-255 (the difference between Windows 1252 and Latin 1), so a simple test for byte/char equality is no longer possible.

This also extends the range of the test from 32-255 to 0-255 for binary and 0-500 for text. This additional range became possible with the Unicode FMSLogo.

Remove an extra space from a comment

Rename ecma_* functions:
ecma_set -> vbar_set
ecma_clear -> vbar_clear
ecma_get -> is_vbarred

I think "ecma" was to indicate this shifting was to support Western European characters, although even then, as a prefix it doesn't make sense because it shifted special characters to ASCII control characters. Now that FMSLogo supports Unicode, the using the "ecma" prefix is even sillier.

"vbar" better describes what these functions are doing. And "is_vbarred" was preferred to "vbar_get" because the function is a predicate and doesn't "get" anything.

Refactoring: Change g_SpecialCharacters to an array instead of string literal
This removes the need to subtract one to remove the NUL terminator.
It also removes the temptation of using wscchr(), which includes the NUL.

Update a bug number referenced in a comment.
When Dice bought SourceForge, they re-numbered all of the bugs.

Cleanup: Remove unneeded LOCAL in READWORDTEST.UICANCEL
This was probably a copy&paste error.