Update the version for apple packages on the release branch (pytorch#…

…7652)

Update

[ghstack-poisoned]

Remove quantize_io flag again.

The quantize io flag was removed in a prio commit.

Signed-off-by: Per Åstrand <per.astrand@arm.com>
Change-Id: Id10d6502fe4b39d453fc04d758666575e833ce19

[Arm backend] Refactor runner_utils, removing RunnerUtil

As the execution of models on reference_model
or on FVP has simplified, there is no longer need
for a class to handle the execution.
Instead, running a model should should only be a
function call with the model, inputs and some metadata.
This makes it easier to reuse runner_utils.

Change-Id: Ib6addbdaa2fa492c356e92ad1b95c2dac2885b9d

Merge branch 'pytorch:main' into change-969691

Update on "use-pt-pinned-commit for test-arm-{backend,reference}-dele…

…gation"

Without this, these builds don't respect the torchgen pinned commit and thus fail with pytorch#7546.

Differential Revision: [D67996459](https://our.internmc.facebook.com/intern/diff/D67996459/)

[ghstack-poisoned]

Merge branch 'pytorch:main' into dev/refactor_unittests

Update on "Use c10 version of half/bfloat16 in executorch"

Accomplished by importing relevant files from c10 into
executorch/runtime/core/portable_type/c10, and then using `using` in
the top-level ExecuTorch headers. This approach should keep the
ExecuTorch build hermetic for embedded use cases. In the future, we
should add a CI job to ensure the c10 files stay identical to the
PyTorch ones.

Differential Revision: [D66106969](https://our.internmc.facebook.com/intern/diff/D66106969/)

[ghstack-poisoned]

Trigger wheel builds when modifying setup.py or pyproject.toml

These two files are the core of the wheel configs, so it makes sense
to try building wheels when they change.

Sort the path entries now that the list is getting longer.

Add TagUnquantizedNodesPass

For models with operations that are not quantized, this pass keeps unquantized operators on the CPU. For example, the deit-tiny-patch16-224 network has an unquantized scaled_dot_product_attention operation. When compiling to Vela, invalid argument errors occur because unquantized operations are offloaded to the NPU. This pass is designed to solve this problem.