[executorch][native] Add Scalar value type to the in-memory IR

[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

Add Scalar (backends/native/runtime/graph/Scalar.{h,cpp}), a concrete scalar
value — a std::variant<int64_t, double, bool>, analogous to c10::Scalar:
one alternative per domain (integral, floating point, boolean), each stored at
its widest type, so a narrower value is stored exactly and to<T>() narrows it
back on read.

Construct implicitly from int / int64_t / double / bool; read via strict
accessors (to_int / to_double / to_bool, which throw on a tag mismatch) or
the promoting to<T>() template (a std::visit that static_casts whichever
alternative is live, like c10::Scalar::to<T>()). Plus is_*() predicates. No
"none" state — the graph's Value owns that.

Every pointer converts to bool, so the implicit constructors would otherwise
let Scalar s = some_ptr; compile and quietly produce a Bool. A deleted
Scalar(T*) template rejects that at the call site while leaving the numeric
constructors implicit.

to_string(const Scalar&) joins the IR printer in utils/Print.{h,cpp}. It
renders a double through the shared ptn::format_double
(graph/StringFormat.h) rather than std::to_string, whose fixed six-decimal
format prints 1e-300 as "0.000000" and 1e300 as 312 digits.

The constructors and the is_*() / to<T>() readers are constexpr,
so a Scalar can be built and inspected in a constant-evaluated context.

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: D114396762

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22276

• 📄 Preview Python docs built from this PR

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

✅ You can merge normally! (1 Unrelated Failure)

As of commit cd0319a with merge base a5f15b5 ():

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-qnn-testsuite-linux / test-backend-linux (qnn, models) / linux-job (gh) (trunk failure)
  The runner has received a shutdown signal. This can happen when the runner service is stopped, or a manually started runner is canceled.

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.