[Bug] Cannot differentiate any coefficients in new opmath expvals in a jitted qnode

[Qottmann]

The problem is we are hitting the is_hermitian check in pennylane/measurements/expval.py which does not work for abstract tracers.

Same problem for qml.sum(qml.s_prod(x, X(0))) or qml.prod(qml.s_prod(x, X(0)))

dev = qml.device("default.qubit")

@jax.jit
@qml.qnode(dev, interface="jax")
def qnode(x):
    return qml.expval(qml.s_prod(x, X(0)))

x = jnp.array([0.5])
jax.grad(qnode)(x)

File ~/Xanadu/pennylane/pennylane/measurements/expval.py:74, in expval(op)
     68 if isinstance(op, qml.Identity) and len(op.wires) == 0:
     69     # temporary solution to merge https://github.com/PennyLaneAI/pennylane/pull/5106
     70     raise NotImplementedError(
     71         "Expectation values of qml.Identity() without wires are currently not allowed."
     72     )
---> 74 if not op.is_hermitian:
     75     warnings.warn(f"{op.name} might not be hermitian.")
     77 return ExpectationMP(obs=op)

File ~/Xanadu/pennylane/pennylane/ops/op_math/sprod.py:201, in SProd.is_hermitian(self)
    197 @property
    198 def is_hermitian(self):
    199     """If the base operator is hermitian and the scalar is real,
    200     then the scalar product operator is hermitian."""
--> 201     return self.base.is_hermitian and not qml.math.iscomplex(self.scalar)

    [... skipping hidden 1 frame]

File ~/anaconda3/envs/pennylane311/lib/python3.11/site-packages/jax/_src/core.py:1510, in concretization_function_error.<locals>.error(self, arg)
   1509 def error(self, arg):
-> 1510   raise TracerBoolConversionError(arg)

TracerBoolConversionError: Attempted boolean conversion of traced array with shape bool[1]..
The error occurred while tracing the function qnode at /var/folders/yv/h7q98p6d2td8vzzbdszwc5l40000gq/T/ipykernel_49113/2688043293.py:3 for jit. This concrete value was not available in Python because it depends on the value of the argument x.
See https://jax.readthedocs.io/en/latest/errors.html#jax.errors.TracerBoolConversionError