Support for Unicode Roman Numerals (e.g. Ⅳ, Ⅷ, etc.)

[vanviethieuanh]

Hi!

As what we discussed in this discussion I implemented:

• Unicode support for fn from_char(c: char) -> Result<Vec<Digit>>.
• Split the feature called archaic because characters like ↈ is rarely used.
• Add more doc tests and change 2 of them mostly because we now return Vec<Digit>.

As suggestion:

Would have to look at adding similar support to the Roman conversions as well

Now Roman::from_str could parse mixing string like XⅦ. However, because Unicode characters like Ⅶ is more than 1 byte using from_byte have to become from_char and result of map become 2D Vec instead. As a result, it ~5 times slower than the original (tested on mmmdcccxciii case) and we have to change most of the code in Roman::from_str.

I tried to use both collect.iter.flatten and for with extend to avoid changing your code. But with this, performance drop even more (~7-10 times slower). This is all I could do to preserve your logic, please let me know if you want to revert what I changed in Roman