If you want your base256 encoding to be as efficient as possible you should use every available printable ascii character. In UTF-8 Ascii character is represented in 1 byte while anything beyond ascii is 2 bytes, so using all 94 (without space) or 95 (with space) and only using Unicode characters in the alphabet after the from 95-255 or 96-255..
Yes, using more ASCII characters would be more efficient (1 byte vs 2 bytes). However, I decided against it for 2 reasons:
Just small data is typically represented in Base256, e.g. 8/16/32/64/128 bytes for passwords/hashes/etc. Therefore, efficiency is not top priority.
Please note that non-terminal characters have been used only for Base256 to support double-clicking for copy&paste. With terminal characters (+-/*.,%&=?...) this would be much more complicated and error-prone. Just think of period or commata at the end of Base256 data.
If you want your base256 encoding to be as efficient as possible you should use every available printable ascii character. In UTF-8 Ascii character is represented in 1 byte while anything beyond ascii is 2 bytes, so using all 94 (without space) or 95 (with space) and only using Unicode characters in the alphabet after the from 95-255 or 96-255..