Move to ExecuTorch 1.5.1

[MatthiasHertelArm]

Change

Moves the pack and the Python exporter to ExecuTorch 1.5.1 together, following the steps in documentation/pack-provenance.md:

Pin	Before	After
PyTorch::ExecuTorch pack (csolution, cproject)	1.4.1	1.5.1
executorch	1.4.1	1.5.1
torch	2.13.0	2.14.0
torchao	0.18.0	0.18.0
ethos-u-vela, tosa-tools, flatbuffers, ml_dtypes	unchanged	unchanged

torch 2.14.0 is what torch_pin.py names at the v1.5.1 tag; the 1.5.1 wheel requires torchao>=0.18.0,<0.19. The Arm backend's requirements-arm-tosa.txt and requirements-arm-ethos-u.txt pin the same versions as in 1.4.1, so requirements-arm-tosa.txt here does not change.

The regenerated ai_layer/ and the RTE context headers name the new pack version. The README, pack-provenance.md and the comment in setup_venv.py follow; pack-provenance.md now also points at torch_pin.py, where the torch version lives since 1.5.

Checked

• The pack is in the public index (cbuild setup --packs finds it) and the wheels resolve from PyPI.
• The export runs with the new pins; the program is 8,784 bytes instead of 8,832, with the same NPU operations.
• The 1.5.1 pack and exporter build with AC6 and pass on the Corstone-320 FVP in a derived project with a larger model.
• CI builds this branch with AC6, GCC and CLANG and runs each image on the FVP: all three pass and print the same logits as with 1.4.1. The venv check passes on Linux, macOS and Windows.

Note for merging

This PR regenerates ai_layer/model_pte.c. Any other open PR that regenerates the layer needs it regenerated again after this one merges.