Cortex-M: size scratch buffers in a dedicated pass

[rascani]

Move CMSIS-NN scratch allocation sizing out of AtenToCortexMPass into InitializeScratchBuffersPass. Run it immediately after dialect lowering in both default pipelines. This lets later graph rewrites, such as padding fusion, run before sizing without embedding them in dialect conversion.

Preserve target-specific sizing, metadata updates and validation of trailing allocation nodes. Recompile after changing allocation arguments. Explicit custom pipelines that perform dialect lowering must include the sizing pass before memory planning. The extraction is independently reviewed; existing convolution, pooling, layout and DS-CNN dialect tests exercise both pipelines.

Authored with AI assistance from Codex.

[rascani]

Stack from ghstack (oldest at bottom):

• Remove unused exported tensor constants after Cortex-M lowering #22862
• Cortex-M: fuse SAME padding into quantized convolutions #22861
• -> Cortex-M: size scratch buffers in a dedicated pass #22860
• Cortex-M: avoid redundant convolution and pooling work #22859

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22860

• 📄 Preview Python docs built from this PR

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

✅ No Failures

As of commit 57305bb with merge base 026ca3f ():
💚 Looks good so far! There are no failures yet. 💚

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