(uxnasm) only ignore [ or ] if it is a whole token

Currently, tokens beginning with a [ or ] character are completely
ignored, which forbids a macro from beginning with these characters.
Specifically, a macro can be declared eg. as `%[x { ... }` but cannot be
dereferenced as `[x`.
This patch only ignores these tokens if they have a length of 1;
otherwise the switch falls through to the default case.

src/uxnasm.c

@@ -343,8 +343,8 @@ parse(char *w, FILE *f)
 		while((c = w[++i]))
 			if(!writebyte(c)) return 0;
 		break;
-	case '[': break; /* ignored */
-	case ']': break; /* ignored */
+	case '[': if (slen(w) == 1) break; /* else FALLTHROUGH */
+	case ']': if (slen(w) == 1) break; /* else FALLTHROUGH */
 	default:
 		/* opcode */
 		if(findopcode(w) || scmp(w, "BRK", 4)) {