rna-transcription approach: a few improvements

[petrem]

• Renamed chr to char in the snippets because chr is a built-in function and although shadowing it in this case may not be a problem, it is still a bad practice when it can be avoided.
• The dictionary-join approach mentions list comprehensions, but instead it uses a generator expression. Replaced this in the explanation and expanded to give the list comprehension based implementation along with a brief comparison.
• The overview mentions one approach is four times faster. In a brief comparison, it varies from 2.5x for a very short string and up to 60x faster for a 10^6 long one. Probably not worth going into the details, but 4x is just innacurate.

I noticed that the maketrans based approach mentions the translation is done between ASCII ordinal values. This happens to be true because the characters involved are ASCII letters, but the functions deal with unicode ordinals.

Do you think this is worth mentioning?