refactor(python): rework the bindings with latest MLIR version

- Go through CAPI for python bindings
- Consuming LLVM errors in CAPI: fixes previous issue which made this
  impossible in the python bindings

compiler/test_python.py

@@ -11,14 +11,29 @@ def main():
         zamalang.register_dialects(ctx)
 
         module = Module.create()
-        eint16 = hlfhe.EncryptedIntegerType.get(ctx, 16)
+        eint6 = hlfhe.EncryptedIntegerType.get(ctx, 6)
         with InsertionPoint(module.body):
-            func_types = [RankedTensorType.get((10, 10), eint16) for _ in range(2)]
+            func_types = [MemRefType.get((10, 10), eint6) for _ in range(2)]
             @builtin.FuncOp.from_py_func(*func_types)
-            def fhe_circuit(*arg):
+            def main(*arg):
                 return arg[0]
 
         print(module)
+        m = """
+        func @main(%arg0: !HLFHE.eint<2>) -> !HLFHE.eint<2> {
+            %0 = constant 1 : i3
+            %1 = "HLFHE.add_eint_int"(%arg0, %0): (!HLFHE.eint<2>, i3) -> (!HLFHE.eint<2>)
+            return %1: !HLFHE.eint<2>
+        }"""
+        ## Working when HFLFHE and MLIR aren't linked
+        zamalang.compiler.round_trip("module{}")
+        zamalang.compiler.round_trip(str(module))
+        ## END OF WORKING
+        ## Doesn't work yet for both modules
+        engine = zamalang.CompilerEngine()
+        engine.compile_fhe(m)
+        # engine.compile_fhe(str(module))
+        print(engine.run(2))
 
 
 if __name__ == "__main__":