Core operator set, scope and coordination

[anssiko]

The WG has been discussing core operator set webmachinelearning/webnn#573 and is analyzing what compositional fundamentals are needed. As part of this effort, a review of e.g. MLIR Linalg, PyTorch Prims IR, TOSA, StableHLO, is being conducted.

The WG charter scope section is abstract enough to not warrant a revision to allow this work to happen. Notably, the following statement seems future-proof:

This Working Group puts priority on building blocks required by well-known model architectures [...]

So I'm not proposing any normative changes to the scope section.

There are a few informative things to check in this context, however:

• Any updates to the informative list of major platform APIs? Currently reads:

  The APIs in scope of this group are not tied to any particular platform and are implementable on top of existing major platform APIs, such as Android Neural Networks API, Windows DirectML, and macOS/iOS Metal Performance Shaders and Basic Neural Network Subroutines.

• Only StableHLO out of many is currently mentioned, see external coordination. This seems unbalanced, given the group is also looking at other sets. Does it help the reader to mention other sets, or is it better to remove the lone StableHLO reference to simplify?