Research: ground Semantic Projection Networks in prior art

[cemphlvn]

Summary

Adds a non-normative literature-grounding note for the proposed Semantic Projection Networks (SPNs) research direction.

The note narrows the novelty boundary after reviewing adjacent work in:

• bidirectional transformations and symmetric lenses;
• multiary delta lenses and networks of models;
• MLIR / heterogeneous IR infrastructure;
• semantics-preserving compilation;
• equality saturation;
• Hopfield dynamics and cortical communication subspaces as inspiration only.

The strongest correction is that network-shaped interoperability itself is prior art. The candidate OSIL contribution is instead framed as evidence-carrying semantic transport across heterogeneous computational representation networks, with preservation/loss contracts that compose over paths and constrain search.

It also adds six research questions and immediate falsifiable experiments.

Search-status disclosure

The requested Consensus search could not run because the connected account's monthly search quota was exhausted. The note says this explicitly and uses primary/publisher/bibliographic sources from an independent academic-web search instead. It should not be described as a Consensus systematic review.

MIT / DCO sign-off included in the commit.