Cortex_m backend: Add IO quantizers + tests of non rescaling ops

[AdrianLundell]

A number of ops only handles shape/meta-data without changing the dynamic range. In these cases, no rescaling needs to be performed and the int8 portable_ops kernel can be used directly.

A new test is added to ensure this behaviour, as well as a test showing how operators which does change the dynamic range (SUB) are not supported.

To support quantization of graphs with no-rescale ops in the beginning/ end of the graph, two new quantizers InputQuantizer and OutputQuantizer are introduced. By explicitly stating the dtpye of the input/output, no-rescale ops inherit dtypes from them as with any other op.

cc @freddan80 @per @zingo @oscarandersson8218 @digantdesai

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15590

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (5 Unrelated Failures)

As of commit 094b370 with merge base 993254c ():

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• pull / test-openvino-linux / linux-job (gh) (similar failure)
  ##[error]The operation was canceled.

BROKEN TRUNK - The following jobs failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-binary-size-linux-gcc / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / test-setup-linux-gcc / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest / windows / windows-job (gh) (trunk failure)
  backends/xnnpack/test/ops/test_conv1d.py::TestConv1d::test_qs8_conv1d_batchnorm_seq
• pull / unittest-editable / windows / windows-job (gh) (trunk failure)
  backends/xnnpack/test/ops/test_conv1d.py::TestConv1d::test_qs8_conv1d_batchnorm_seq

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[psiddh]

Is this the idea in general ?

// Current state:
input(float32) → ops → output(float32)

// Annotate (This PR):
input(float32) 📋"Annotate Q!" → ops → 📋"Annotate DQ!" output(float32)

// What later passes will do:
input(float32) → [Q] → ops(int8) → [DQ] → output(float32)
↑---------------------------------↑
"OK, inserting ------------- ---------"OK, inserting
quantize!" -------------------------- "dequantize!"