Make it clear(er) that Scrubcsv favors snake_case over CamelCase in header

[ngirard]

It seems like Scrubcsv favors snake_case in the header, since A B is transformed into a_b.
I'm fine with that, although it's not part of RFC 4180.
I'd suggest making it clear(er) by

• mentioning it in the README;
• adding a dedicated test in the tests.rs file, similar to the one below;
• making it optional with a dedicated command-line option;
• being more consistent by replacing CamelCase with camel_case

My test:

echo "Aa,A B,CamelCase"|scrubcsv --clean-column-names
aa,a_b,camelcase
1 rows (0 bad) in 0.00 seconds, 214.33 KiB/sec

(I'd expect to get aa,a_b,camel_case instead)