[Feature Request] Missing optimization of DequantizeLinear ∘ Flatten ∘ QuantizeLinear?

[mcollinswisc]

Discussed in #21167

^{Originally posted by mcollinswisc June 25, 2024}
It looks like ONNXRuntime will optimize DequantizeLinear ∘ Reshape ∘ QuantizeLinear to only the Reshape, eliminating the quantization/de-quantization, if the scales & zero points are the same.

However, an equivalent Flatten is not optimized. Is this likely to be just a missing optimization, or is there some reason the qdq would be preserved in this case?

Tested out in:
https://gist.github.com/mcollinswisc/d1cd9d13b4e5fbad01c75dca5c9ca576
with ONNXRuntime 1.18.0

[skottmckay]

Should be possible to add to this list given the ONNX spec for Flatten allows 8-bit integers:

onnxruntime/onnxruntime/core/optimizer/qdq_transformer/selectors_actions/qdq_selector_action_transformer.cc

Lines 63 to 68 in 0f1f3b7

	qdq_selector_action_registry.RegisterSelectorAndAction(drop_action_name,
	{{"Gather", {}},
	{"Reshape", {}},
	{"Transpose", {}},
	{"Squeeze", {}},
	{"Unsqueeze", {}}},