readChar and printChar are more UTF-8 aware

[privat]

Currently, input and output dealing with Unicode is inconsistent.

Some operation assumes UTF-8 where chars are just bytes and 'é' is `c3 a9, some assume internal UTF-16, some other assumptions are platform dependent (because it is the bad default of many Java methods).

Previously:

• printChar prints the lowest bytes as a character, so if a0 is a9c3 then c3 is understood as a code-point, so Ã is printed.
• readChar depends on the mode, in terminal and GUI mode, it seems that utf-8 is used: bytes are read as is. In popup mode, the code-point of the first character read is used, that might even set a0 to a value larger than 0xff or even negative

This PR makes both operations more consistent with a common UFT-8 encoding

• readChar read bytes, so if the input is c3 a9, readchar consume and return c3 in all mode and systems
• printChar print the lowest byte as is, since c3 alone is not a valid UTF-8 unicode character, a replacement character is displayed by the GUI.

Benefits:

• Behavior more consistent
• in case of bugs, replacement character likely used instead of some random character, making the behavior more explainable.

Problem that remain:

printChar print each character independent, so even is we printChar c3 then a3, 2 replacement character will be issued instead of a single é.