Add hasInput method to InferenceContext

[gramalingam]

See #4447 for a discussion.

[p-wysocki]

LGTM, I think it's a great addition.

[gramalingam]

@daquexian can you please take a look? This requires the approval from someone in operator SIG. Thanks!

[jcwchen]

Great improvement! Shall we also apply hasInput in

1. https://github.com/onnx/onnx/blob/main/onnx/defs/tensor/utils.cc#L95-L99
2. https://github.com/onnx/onnx/blob/main/onnx/defs/quantization/old.cc#L41

Besides, do you think it is also useful if we can have another common function for ctx.getNumInputs() > x && hasInputShape(ctx, x)? For example: https://github.com/onnx/onnx/blob/main/onnx/defs/nn/old.cc#L70-L77 It can be done by another PR if needed.

[gramalingam]

Great improvement! Shall we also apply hasInput in

1. https://github.com/onnx/onnx/blob/main/onnx/defs/tensor/utils.cc#L95-L99
2. https://github.com/onnx/onnx/blob/main/onnx/defs/quantization/old.cc#L41

Besides, do you think it is also useful if we can have another common function for ctx.getNumInputs() > x && hasInputShape(ctx, x)? For example: https://github.com/onnx/onnx/blob/main/onnx/defs/nn/old.cc#L70-L77 It can be done by another PR if needed.

Added the first two, thanks! As for the next suggestion, hasInputShape already works this way, so we don't need a new function I think. (The extra condition ctx.getNumInputs() > x is not needed.)

[linkerzhang]

Good improvement!

[daquexian]

LGTM :)