feat(pay): upgrade OpenAPI spec to 3.1 and progenitor to 0.13

[chris13524]

Summary

• Updates crates/yttrium/src/pay/openapi.json from OpenAPI 3.0.3 to 3.1.0 (version 2026-02-18).
• Replaces progenitor 0.11.2 (crates.io) with the WalletConnect/progenitor fork at oas3-migration, which adds OpenAPI 3.1 support and includes a header-param Option fix. Depends on Migrate to oas3 (OpenAPI 3.1 support) + header param Option fix WalletConnect/progenitor#1.
• Adapts pay/mod.rs to the new generated types from progenitor 0.13.

Type changes consumed

• PaymentId is now a strict newtype that only impls From<String>, so .id(&payment_id) becomes .id(payment_id.clone()) (or .to_string() when starting from &str).
• CollectData.url is non-optional String; CollectData.schema is serde_json::Value (not Option<Value>).
• Amount.unit is now the generated types::String enum (Iso4217/Caip19 variants); flattened back to the prefixed string form ("iso4217/..." / "caip19/...").
• GetPaymentOptionsResponse.options is now Option<Vec<...>>; unwrapped with .unwrap_or_default().

Known issue

The new openapi.json defines a top-level String schema as a oneOf of two object variants (Iso4217/Caip19), but the runtime format (per the spec example) is a plain prefixed string like "iso4217/USD". Serde deserialization will likely fail at runtime on the unit field. This is a schema authoring issue rather than a progenitor or code issue, but worth flagging — the schema may need to revert unit to a plain string type with a description.

Test plan

• cargo check --features=full --lib --bins -p yttrium passes
• cargo check --features=full --tests passes
• Pay e2e tests (just test-pay-e2e) — requires REOWN_PROJECT_ID
• Verify Kotlin/Swift bindings still generate (just kotlin, just swift)
• Verify runtime behavior of the unit field once the schema issue is resolved

🤖 Generated with Claude Code

[github-actions]

Thank you for your contribution! We ask that you please read and sign our CTA Document before we can accept your contribution. You can sign the CTA simply by posting a Pull Request Comment with the following text:

---

I have read the CTA Document and I hereby sign the CTA

---

Chris Smith seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[claude]

Claude finished @chris13524's task in 3m 44s —— View job

---

PR Review: feat(pay): upgrade OpenAPI spec to 3.1 and progenitor to 0.13

• Gather context
• Read changed files
• Run license compliance check
• Run breaking changes check
• Analyze and post review

---

License compliance: No new license issues. progenitor was already MPL-2.0 at 0.11.2; only truly new dep is oas3 (MIT). ✅

Breaking changes: Public/FFI surface is unchanged — CollectDataAction, PayAmount, PaymentOptionsResponse all keep the same types via the adapter layer. ✅

---

Found 3 issue(s)

Issue 1: Runtime deserialization will fail for Amount.unit field

ID: mod-amount-unit-deser-fail-9c2a
File: crates/yttrium/src/pay/mod.rs:677
Severity: CRITICAL
Category: bug

Context:

• Pattern: types::String is a progenitor-generated enum expecting {"Iso4217": "USD"} or {"Caip19": "..."} on the wire, but the API sends the plain prefixed string "iso4217/USD".
• Risk: Every call to get_payment_options / get_required_payment_actions that returns an Amount will panic/error at the serde deserialization step, not the conversion step. The From<types::Amount> conversion at line 677 is correct — the bug is upstream of it, when progenitor deserializes the HTTP response body.
• Impact: All payment option fetching is broken at runtime. The PR itself flags this as a "Known issue", but it should be resolved before merging.
• Trigger: Any live API call returning an Amount object.

Recommendation: Fix the String schema in openapi.json to use a plain string type (matching the actual wire format) rather than the oneOf object variants. The description and example can document the prefix convention:

"String": {
  "type": "string",
  "description": "Currency unit: 'iso4217/USD' for fiat, 'caip19/eip155:1/erc20:0x...' for tokens",
  "example": "iso4217/USD"
}

Then types::String becomes a plain String and the match in mod.rs:677 simplifies back to unit: a.unit.

---

Issue 2: Git-pinned fork dependency for progenitor

ID: cargo-progenitor-git-pin-b3f1
File: Cargo.toml:53
Severity: HIGH
Category: maintainability

Context:

• Pattern: progenitor and progenitor-client are sourced from a WalletConnect GitHub fork pinned to commit 664b34bb321d47559752059b7fe0e07fd96419ee instead of a crates.io release.
• Risk: If that commit is force-pushed or the fork is deleted, the build breaks with no recoverable path. The fork's oas3-migration branch also depends on a PR (WalletConnect/progenitor#1) that may not be merged upstream.
• Impact: Build instability for all contributors and CI; cannot be audited by cargo audit.
• Trigger: Any cargo build after the fork's git history is modified.

Recommendation: Either publish the fork to crates.io, or add a comment documenting the expected lifetime of this dependency and a tracking issue to upstream the OAS 3.1 support to the official progenitor crate.

---

Issue 3: Silent error suppression in CollectDataAction schema serialization

ID: mod-collectdata-schema-silent-err-d4e8
File: crates/yttrium/src/pay/mod.rs:616
Severity: LOW
Category: code_quality

Context:

• Pattern: serde_json::to_string(&c.schema).ok() silently drops a serialization error, yielding schema: None.
• Risk: serde_json::Value serialization is practically infallible, but the inconsistency with line 552 (unwrap_or_else(|e| { ... log error ... })) means if it ever does fail, the schema is silently dropped without any log signal.
• Impact: Hard-to-diagnose data loss in the webview collect flow.

Recommendation: Match the pattern used at line 552:

schema: serde_json::to_string(&c.schema)
    .map_err(|e| pay_debug!("schema serialize error: {e}"))
    .ok(),

[chris13524]

I have read the CTA Document and I hereby sign the CTA