Re-land Python buffer inputs after internal CI fixes (pytorch#23535)

## Summary

Corrected re-land of pytorch#23088 after pytorch#23532.

- accept NumPy arrays and other CPU Python buffer providers directly
- retain borrowed storage for the runtime input lifetime
- handle empty and read-only buffers safely
- reject resized borrowed inputs before execution
- preserve all existing Torch-input behavior
- avoid importing torch on buffer-only call paths

## Internal CI fixes

- removed unsupported ChannelsLast3d bridging from the portable Torch
path
- restored legacy RuntimeError behavior for unsupported Torch layouts
- kept the zero-source tensor-copy behavior unchanged
- added regression coverage for buffer-only calls and empty tensor
copies

## Test plan

- Python syntax validation
- internal ATen and portable pybindings targets
- internal standard and custom-library variants
- low-level ATen tensor utility tests

Differential Revision: D123893454

Author: JacobSzwejbka

extension/pybindings/README.md

@@ -28,8 +28,11 @@ CMAKE_ARGS="-DEXECUTORCH_BUILD_VULKAN=ON" ./install_executorch.sh
 ## Classes
 ### ExecuTorchModule
 - `plan_execute()`: Plan and execute.
-- `run_method()`: Run method.
-- `forward()`: Forward. This takes a pytree-flattend PyTorch-tensor-based input.
+- `run_method()`: Run a method with either PyTorch tensors or objects that
+  implement Python's buffer protocol, such as NumPy arrays. A call must use a
+  single tensor protocol; PyTorch tensors and buffers cannot be mixed.
+- `forward()`: Forward. A single buffer may be passed directly; multiple
+  inputs are passed as a flat sequence.
 - `has_etdump()`: Check if etdump is available.
 - `write_etdump_result_to_file()`: Write etdump result to a file.
 - `__call__()`: Call method.