Fix symbol parsing

[comnik]

Fix extracted from one of Nate's old PRs

BUG: SYMBOL lexer regex has incorrect character class

  Location: grammar.y:46 (propagates to all generated SDKs)

  Problem: The regex [a-zA-Z0-9_./#-] is incorrect. Inside a character class, - between / and ] is ambiguous — in some regex engines it's interpreted as a range
from
   # (0x23) to - (0x2D), which unintentionally includes characters like $, %, &, ', (, ), *, +, ,.

  Fix: Move the - to the end of the character class so it's treated as a literal hyphen:

  [a-zA-Z0-9_.#/-]*

[minsungc]

LGTM, thanks!