There's something weird going on here. If you play 21. 15-33 instead of 15-42 then Scan will indicate that the value of the position is 0 and that the best move is 16-27, no matter how long you let it analyse – but after playing 16-27, the score is +3 again. Looks like a bug to me.
For some unknown reason Scan doesn't play the right moves. After 7. ... 27-31 ? the position is lost for Black (Dragon Draughts database). Maybe Scan doesn't use endgame database in lidraughts ? Anyway, Black gets more defense by playing 7. ... 13-18! then it's much more difficult for White to win. For example : 8. 19-14 27-31 9. 14-9 (48-42 25-30! 24x35 15-20! 14x25 18-22 and then 22-27-32 =) 31-37 10. 9-4 18-23 11. 4-36 23-28 12. 36-47 28-33 13. 47x29 37-41 14. 29-47 41-46 15. 47-33 46-23 16. 33x6 23x45 17. 6-1 W+ (database)
16-27 repeats a position. During Scan's search, a repetition immediately ends the game, and is scored as a draw. This makes search more efficient, by avoiding searching moves sequences that have cycles in them (which can never improve the result).
To add to @Faraday's question, when you run Scan in the browser on the lidraughts analysis boards, no tablebases are used (too large to download and run locally). When requesting server-side analysis Scan does use 5-piece tablebases.