Major Description
Following OAK-1007 where I switched to a ClassicAnalizer, I realized that it introduced some subtle changes in tokenization behavior.
For example there's a twist if the token contains a number.
From the ClassicTokenizer api:
Splits words at hyphens, unless there's a number in the token, in which case the whole token is interpreted as a product number and is not split.
this means that a path token could be split either in 2 tokens if it has no numbers:
/parent/child => 'parent', 'child'
or just one if it has numbers:
/p12345/p23456 => '/p12345/p23456'
Also, I'd like to split alphanumeric tokens on '_' and on '.' as well.