[rewriter] Remove redundant op.Slice and op.ScatterND

[titaiwangms]

Fixes https://github.com/microsoft/onnx-converters-private/issues/270

It is observed that ExportedProgram could generates aten::slice.Tensor and aten::slice_scatter.default that slice nothing:

The slices would result in redundant op.Slice ops in ONNX graph that does nothing, and op.ScatterND that basically replaces the whole input to updates, which takes a lot of time in inference.

This rule set recognizes the redundant slices by checking if the following requirements are met:
(1) starts = 0
(2) ends >= inputs[dim].shape or ends == _INT64_MAX
(3) steps == 1

This rule set recognizes the redundant scatterND by checking if the following requirements are met:
(1) indices has the same length as the first dim of input
(2) indices is from 0 to input.shape[0]
(3)input has the same shape as updates

Benchmark on ghostnet_100 (the original speed up was 0.0256):

	Stat	Speedup	Increase	Med
Suite	Model Name	onnx_dynamo	onnx_dynamo	onnx_dynamo
Timm	ghostnet_100	1.1599	15.986%	1.1580

[codecov]

Codecov Report

Attention: Patch coverage is 74.10072% with 36 lines in your changes missing coverage. Please review.

Project coverage is 75.23%. Comparing base (c13e4fd) to head (7eca5a6).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
onnxscript/rewriter/collapse_slices.py	53.52%	22 Missing and 11 partials ⚠️
onnxscript/rewriter/testing.py	88.88%	1 Missing and 2 partials ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #1925    +/-   ##
========================================
  Coverage   75.23%   75.23%            
========================================
  Files         254      257     +3     
  Lines       27644    27782   +138     
  Branches     3225     3242    +17     
========================================
+ Hits        20797    20901   +104     
- Misses       5877     5900    +23     
- Partials      970      981    +11     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.