Slice behavior wrong with negative step and end = INT_MAX

Describe the bug
ORT's Slice tests assert that Slice(x, starts = [-1], ends = [INT_MAX], steps = [-1]) == Reverse(x).

ONNX spec says:

For slicing to the end of a dimension with unknown size, it is recommended to pass in INT_MAX when slicing forward and INT_MIN when slicing backward.

I haven't yet tested to see if the behavior is correct with INT_MIN, but that should be checked as well.

Urgency
None.

To Reproduce
Build ORT and run onnxruntime_test_all --gtest_filter='SliceTest.Slice1D_ReverseAllAxes_1'.
Currently the test triggers a shape inference failure which results in a warning being logged.

Expected behavior
Slice(x, starts = [-1], ends = [INT_MAX], steps = [-1]) == <empty tensor>
Slice(x, starts = [-1], ends = [INT_MIN], steps = [-1]) == Reverse(x)

Both cases should be tested in ORT and not result in shape inference failures.

Additional context
Add any other context about the problem here. If the issue is about a particular model, please share the model details as well to facilitate debugging.