CoreML InstanceNorm3d fails to load

🐛 Describe the bug

Models with InstanceNorm3d on the Core ML backend fail to load at runtime.

Repro:

import torch

from executorch.backends.apple.coreml.partition import CoreMLPartitioner
from executorch.exir import to_edge_transform_and_lower, EdgeCompileConfig
from executorch.extension.pybindings.portable_lib import _load_for_executorch_from_buffer

class Model(torch.nn.Module):
    def __init__(self):
        super().__init__()
        self.norm = torch.nn.InstanceNorm3d(3)
        
    def forward(self, x):
        return self.norm(x)

model = Model()
inputs = (
    torch.randn(1, 3, 4, 4, 4),
)

eager_outputs = model(*inputs)
print(f"Eager: {eager_outputs.shape} {eager_outputs}")

ep = torch.export.export(model.eval(), inputs)
lowered = to_edge_transform_and_lower(
    ep,
    partitioner=[CoreMLPartitioner()],
    compile_config=EdgeCompileConfig(_check_ir_validity=False)
).to_executorch()

print(ep)
print(lowered.exported_program())

et_model = _load_for_executorch_from_buffer(lowered.buffer)
et_outputs = et_model([*inputs])[0]

et_outputs - eager_outputs

Output:

[ETCoreMLModelCompiler.mm:55] [Core ML]  Failed to compile model, error = Error Domain=com.apple.mlassetio Code=1 "Failed to parse the model specification. Error: Unable to parse ML Program: in operation aten_instance_norm_default_cast_fp16: parameter x[0] has invalid rank 5, expecte$
[backend_delegate.mm:288] [Core ML] Model init failed Failed to compile model, error = Error Domain=com.apple.mlassetio Code=1 "Failed to parse the model specification. Error: Unable to parse ML Program: in operation aten_instance_norm_default_cast_fp16: parameter x[0] has invali$
[coreml_backend_delegate.mm:193] CoreMLBackend: Failed to init the model.
[method.cpp:113] Init failed for backend CoreMLBackend: 0x23

Versions

coremltools version 8.3
executorch commit https://github.com/pytorch/executorch/commit/67b6009d6b3b67eee775c8ed2fe30eae6e0bb65c (Jun 14)

cc @kimishpatel @YifanShenSZ @cymbalrush @metascroy

Start with executorch.backends.apple.coreml.partition.CoreMLPartitioner and run the supplied InstanceNorm3d reproduction using the stated Core ML and ExecuTorch versions. Trace the generated Core ML model around aten_instance_norm_default_cast_fp16 and the reported rank-5 parsing failure. Done means the lowered model loads successfully and its outputs can be compared with eager_outputs.