(refactor): preserve partial struct construct info during merge with FieldVar placeholders

[eytan-starkware]

Summary

Introduces placeholder support in equality analysis for struct fields to handle unknown fields during control flow merges. When merging equality states from different branches, fields that don't have intersection representatives are now represented as placeholders instead of being dropped, preserving partial struct information.

---

Type of change

Please check one:

• Bug fix (fixes incorrect behavior)
• New feature
• Performance improvement
• Documentation change with concrete technical impact
• Style, wording, formatting, or typo-only change

---

Why is this change needed?

Previously, when merging equality states from different control flow branches, struct fields that couldn't be unified (no intersection representative) were completely lost. This caused the analysis to lose valuable structural information about partially-known structs, reducing the effectiveness of equality tracking across complex control flows.

---

What was the behavior or documentation before?

Struct fields in equality analysis were represented only as real variables (VariableId). During merge operations, if a field had no intersection representative between two branches, that field information was discarded entirely, potentially losing the entire struct relationship.

---

What is the behavior or documentation after?

Struct fields are now represented as FieldVar enum that can be either a real variable or a globally unique placeholder. During merges, unknown fields become placeholders, preserving the struct's shape and allowing partial equality information to be maintained. Placeholders are displayed as "?" in debug output and are handled appropriately in all struct operations.

---

Related issue or discussion (if any)

---

Additional context

This change includes comprehensive updates to handle placeholders throughout the equality analysis pipeline, including struct construction/destruction, array operations, and debug formatting. The implementation ensures that placeholders are properly tracked and don't interfere with real variable relationships while preserving as much structural information as possible.

[eytan-starkware]

• (optimization): Using equality analysis for variable forwarding #9874
• (feat): Added an optimization dump for each phase under a special log target #9873
• (feat): merging analysis tests #9872
• (feat): dominator analysis using forward dataflow #9871
• (feat): def site analysis based on forward dataflow #9870
• (refactor): preserve partial struct construct info during merge with FieldVar placeholders #9804 👈 (View in Graphite)
• refactor: unify hashcons reverse maps and class_info into a single ClassInfo struct #9848
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[reviewable-StarkWare]

This change is 

[cursor]

PR Summary

Medium Risk
Changes core equality-analysis merge logic to retain partial StructConstruct relationships using synthetic placeholder fields, which could affect downstream optimizations/diagnostics that rely on this analysis. Risk is moderate due to new placeholder semantics and updated array/struct handling, but scope is limited to the lowering equality analysis and its tests.

Overview
Improves EqualityAnalysis so control-flow merges preserve partial struct/array construct information instead of dropping the whole StructConstruct relation when some fields can’t be intersected.

This introduces FieldVar (real var vs globally-unique placeholder) and threads it through relation tracking, struct construct/destructure, and array pop tracking; merge now emits placeholders for non-intersecting fields (rendered as ? in debug output). Adds a regression test demonstrating a merged struct retaining a shared field while marking the differing field unknown.

^{Reviewed by Cursor Bugbot for commit fae55f3. Bugbot is set up for automated code reviews on this repo. Configure here.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4f7c2c867c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[orizi]

@orizi reviewed all commit messages and made 1 comment.
Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on eytan-starkware and TomerStarkware).

---

a discussion (no related file):
no tests are possible?

[eytan-starkware]

@eytan-starkware made 2 comments.
Reviewable status: 0 of 2 files reviewed, 2 unresolved discussions (waiting on orizi and TomerStarkware).

---

a discussion (no related file):

Previously, orizi wrote…

no tests are possible?

They were lost but now found

[eytan-starkware]

@eytan-starkware resolved 1 discussion.
Reviewable status: 0 of 2 files reviewed, 1 unresolved discussion (waiting on orizi and TomerStarkware).

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit e9939df. Configure here.}

[TomerStarkware]

@TomerStarkware reviewed 2 files and all commit messages, and made 1 comment.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on eytan-starkware and orizi).

[orizi]

@orizi reviewed all commit messages, made 1 comment, and resolved 1 discussion.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on eytan-starkware).

[orizi]

@orizi reviewed 1 file and all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on eytan-starkware).