[executorch][native] Add TensorMeta + ScalarType in-memory IR types

[SS-JIA]

Stack from ghstack (oldest at bottom):

• [executorch][native] Add Method + ValueRole to the in-memory IR #22281
• [executorch][native] Add Graph index arena to the in-memory IR #22280
• [executorch][native] Add Node to the in-memory IR #22279
• [executorch][native] Add Argument tagged union to the in-memory IR #22278
• [executorch][native] Add Value + index-arena ids to the in-memory IR #22277
• [executorch][native] Add Scalar value type to the in-memory IR #22276
• -> [executorch][native] Add TensorMeta + ScalarType in-memory IR types #22275
• [executorch][native] Add native runtime Program reader + DOT visualizer #22274

First concrete in-memory IR value types for the native runtime, under
backends/native/runtime/graph/:

• ScalarType (ScalarType.h) — a standalone, header-only scalar element-type
  enum driven by an X-macro table (PTN_FORALL_SCALAR_TYPES(cpp_type, name, id)). The macro generates the enum (ids pinned to ExecuTorch's ScalarType
  / the native_graph.fbs schema so a deserializer maps the serialized byte
  directly), the k<Name> constants, the forward ScalarType -> C++ type
  trait (ScalarTypeToCppType<N> / cpp_type_t<N>), and the element_size()
  / scalar_type_name() helpers. Half / BFloat16 map to uint16_t as a raw
  storage stand-in.
• TensorMeta (TensorMeta.{h,cpp}) — logical tensor metadata: element type,
  per-dim size ranges (Dim{min, max}), and an advisory dim_order_hint
  (non-prescriptive — engines choose their own physical layout). Helpers:
  ndim(), is_static(), is_contiguous(), numel() (upper-bound extent;
  throws on an unbounded dim), and equality.

Dim's range constructor rejects a range no shape can have -- a negative lower
bound, or a bounded upper bound below it -- so a malformed serialized shape
fails where it enters the IR instead of surfacing later as a wrong numel().
numel() itself checks each product against INT64_MAX before multiplying,
since signed overflow is UB and an overflowed count would silently plan a
smaller buffer than the shape asks for.

• to_string(const TensorMeta&) (utils/Print.{h,cpp}, target graph:print) —
  the first entry in the IR's debug printer. A free function in its own target
  rather than a member, so only a consumer that dumps the IR links the
  formatting code; later diffs extend it as each IR type lands.

Pure std — no ExecuTorch and no flatbuffers dependency. Mirrored into both the
fbcode/ and xplat/ trees to match the native backend layout.

Authored with Claude Code.

Differential Revision: D114396765

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22275

• 📄 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 978a757 with merge base 16d3b9f ():
💚 Looks good so far! There are no failures yet. 💚

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

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[digantdesai]

Review automatically exported from Phabricator review in Meta.