FMSLogo Code

Fix Bug #540; WORD now correctly copies characters after a CHAR 0.
The fix is to replace strncpy with memcpy. This fix also helps the Unicode branch work around an apparent bug in wcsncpy in the glibc on 64-bit GNU/Linux.

Refactoring: change the return type of copy_routine to return void.
The return value was never used.
This makes the string copying logic a little simpler.

Reformatting: Remove a blank line

Add a missing colon after an example intro line.

Merge in changes from trunk to get fix for build break on 64-bit GNU/Linux

Add logic to only write a UTF-8 BOM if the workspace as a non-ASCII character.
This was done for so that newer FMSLogo can save files that older FMSLogo can read, provided that they're entirely English (the common case).

Add some comments

Improve the wording of a comment.

Add a missing ~ to the manual test for the colorizing of lines with continuations.

Rewrite the example for SOUND to follow better Logo style.
- FPUT is algorithmically faster than LPUT
- Having two "make" instructions in the loop generated unneeded garbage.

While LPUT might have been clearer, the fact that there was two instructions, one for the frequency and one for the duration was hard to see. To help clarify what the REPEAT instruction is doing, I added an explanatory sentence which shows the gist of what list is being created. Preserving the ascending tones with LPUT would have made the math unnecessarily complex, so I made them descending.