[BUG] default.tensor errors out with mid circuit measurements

[albi3ro]

Expected behavior

I would expect the defer_measurements transform to be applied prior to decompose.

Actual behavior

Error with traceback below.

Additional information

The solution would simply be adding defer_measurements to the transform program before decompose.

Source code

@qml.qnode(qml.device('default.tensor', wires=2))
def circuit():
    m0 = qml.measure(0)
    qml.cond(m0, qml.RX)(0.1, 0)
    return qml.expval(qml.Z(0))

circuit()

Tracebacks

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[16], line 7
      4     qml.cond(m0, qml.RX)(0.1, 0)
      5     return qml.expval(qml.Z(0))
----> 7 circuit()

File ~/Prog/pennylane/pennylane/workflow/qnode.py:995, in QNode.__call__(self, *args, **kwargs)
    993 if qml.capture.enabled():
    994     return qml.capture.qnode_call(self, *args, **kwargs)
--> 995 return self._impl_call(*args, **kwargs)

File ~/Prog/pennylane/pennylane/workflow/qnode.py:985, in QNode._impl_call(self, *args, **kwargs)
    982     self._interface = interface
    984 try:
--> 985     res = self._execution_component(args, kwargs)
    986 finally:
    987     if old_interface == "auto":

File ~/Prog/pennylane/pennylane/workflow/qnode.py:943, in QNode._execution_component(self, args, kwargs)
    940 interface = None if self.interface == "numpy" else self.interface
    942 # pylint: disable=unexpected-keyword-arg
--> 943 res = qml.execute(
    944     (self._tape,),
    945     device=self.device,
    946     gradient_fn=gradient_fn,
    947     interface=interface,
    948     transform_program=full_transform_program,
    949     inner_transform=inner_transform_program,
    950     config=config,
    951     gradient_kwargs=gradient_kwargs,
    952     **execute_kwargs,
    953 )
    954 res = res[0]
    956 # convert result to the interface in case the qfunc has no parameters

File ~/Prog/pennylane/pennylane/workflow/execution.py:528, in execute(tapes, device, gradient_fn, interface, transform_program, inner_transform, config, grad_on_execution, gradient_kwargs, cache, cachesize, max_diff, device_vjp, mcm_config)
    526 # Exiting early if we do not need to deal with an interface boundary
    527 if no_interface_boundary_required:
--> 528     results = inner_execute(tapes)
    529     return post_processing(results)
    531 if (
    532     device_vjp
    533     and getattr(device, "short_name", "") in ("lightning.gpu", "lightning.kokkos")
    534     and interface in jpc_interfaces
    535 ):  # pragma: no cover

File ~/Prog/pennylane/pennylane/workflow/execution.py:204, in _make_inner_execute.<locals>.inner_execute(tapes, **_)
    201 if cache is not None:
    202     transform_program.add_transform(_cache_transform, cache=cache)
--> 204 transformed_tapes, transform_post_processing = transform_program(tapes)
    206 if transformed_tapes:
    207     results = device.execute(transformed_tapes, execution_config=execution_config)

File ~/Prog/pennylane/pennylane/transforms/core/transform_program.py:515, in TransformProgram.__call__(self, tapes)
    513 if self._argnums is not None and self._argnums[i] is not None:
    514     tape.trainable_params = self._argnums[i][j]
--> 515 new_tapes, fn = transform(tape, *targs, **tkwargs)
    516 execution_tapes.extend(new_tapes)
    518 fns.append(fn)

File ~/Prog/pennylane/pennylane/devices/preprocess.py:369, in decompose(tape, stopping_condition, stopping_condition_shots, skip_initial_state_prep, decomposer, max_expansion, name, error)
    366     return (tape,), null_postprocessing
    367 try:
--> 369     new_ops = [
    370         final_op
    371         for op in tape.operations[len(prep_op) :]
    372         for final_op in _operator_decomposition_gen(
    373             op,
    374             stopping_condition,
    375             decomposer=decomposer,
    376             max_expansion=max_expansion,
    377             name=name,
    378             error=error,
    379         )
    380     ]
    381 except RecursionError as e:
    382     raise error(
    383         "Reached recursion limit trying to decompose operations. "
    384         "Operator decomposition may have entered an infinite loop."
    385     ) from e

File ~/Prog/pennylane/pennylane/devices/preprocess.py:64, in _operator_decomposition_gen(op, acceptance_function, decomposer, max_expansion, current_depth, name, error)
     62 else:
     63     try:
---> 64         decomp = decomposer(op)
     65         current_depth += 1
     66     except qml.operation.DecompositionUndefinedError as e:

File ~/Prog/pennylane/pennylane/devices/preprocess.py:355, in decompose.<locals>.decomposer(op)
    354 def decomposer(op):
--> 355     return op.decomposition()

AttributeError: 'MidMeasureMP' object has no attribute 'decomposition'

System information

master.

Existing GitHub issues

• I have searched existing GitHub issues to make sure the issue does not already exist.