Improve Lexeme hashing of punctuation

[raineorshine]

Distinguish the characters -+/:., next to numbers:

• 500 ≠ -500
• 1020 ≠ 10–11
• 12 ≠ 1/2
• 51 ≠ 5:1
• 34 ≠ 3 & 4

(Continue to hash match those adjoining letters, e.g. sense-making and sensemaking)

Distinguish - at whitespace or beginning/end of line:

• A ≠ A-
• Points: 500 ≠ Points: -500

Distinguish @#<>:

• @foo ≠ foo
• #bar ≠ bar

Match &, +, and and:

• one & two = one + two = one and two

Match ö and o:

• Lacoön = Lacoon

[raineorshine]

Fixed in:

a3a5f4cbda normalizeThought: Ignore diacritics. 0d8929d603 normalizeThought: Transform ampersand to 'and'. 10b009f5b0 normalizeThought: Always distinguish hashtags and mentions. af13251894 normalizeThought: Boolean expressions and ending punctuation. 0ea8506644 normalizeThought: Remove colons preceding whitespace. 9be85568fb normalizeThought: Selectively remove hyphens and dashes. d82b28f6c8 Add normalizeThought tests.