Delete evaluator.py shim: all imports go directly to bootstrapped sx_ref.py

EvalError moved to types.py. All 27 files updated to import eval_expr,
trampoline, call_lambda, etc. directly from shared.sx.ref.sx_ref instead
of through the evaluator.py indirection layer. 320/320 spec tests pass.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

shared/sx/tests/test_parity.py

@@ -18,7 +18,7 @@ from shared.sx.types import Symbol, Keyword, Lambda, Component, Macro, NIL
 
 def hw_eval(text, env=None):
     """Evaluate via hand-written evaluator.py."""
-    from shared.sx.evaluator import evaluate as _evaluate, EvalError
+    from shared.sx.ref.sx_ref import evaluate as _evaluate
     if env is None:
         env = {}
     return _evaluate(parse(text), env)
@@ -50,7 +50,7 @@ def ref_render(text, env=None):
 
 def hw_eval_multi(text, env=None):
     """Evaluate multiple expressions (e.g. defines then call)."""
-    from shared.sx.evaluator import evaluate as _evaluate
+    from shared.sx.ref.sx_ref import evaluate as _evaluate
     if env is None:
         env = {}
     result = None
@@ -736,7 +736,7 @@ class TestParityDeps:
 
 class TestParityErrors:
     def test_undefined_symbol(self):
-        from shared.sx.evaluator import EvalError as HwError
+        from shared.sx.types import EvalError as HwError
         from shared.sx.ref.sx_ref import EvalError as RefError
         with pytest.raises(HwError):
             hw_eval("undefined_var")