Utilities shared by the Sifter lexer/parser.

Snakecase rules per segment (segments are dot-separated):

• Insert _ only at lower|digit → UPPER boundaries (e.g. firstName → first_name, id2X → id2_x)
• No acronym splitting (e.g. APIKey → apikey, not api_key)
• Map - and space to _, collapse consecutive _, lowercase everything
• Apply independently to each dot-separated segment (dots preserved)