fix: surface locked Trezor as DeviceBusy via trezor-connect-rs 0.3.4

[coreyphillips]

Context

bitkit-ios #613 (Android counterpart bitkit-android #1030) reports that connecting to a locked Trezor keeps the device busy: the transport was churning in a close -> wait 2000ms -> reopen -> retry loop and ultimately failing with a locked-device handshake error, instead of backing off and letting the user unlock.

The upstream fix landed in trezor-connect-rs 0.3.4, which:

• Adds a typed ThpError::DeviceLocked variant.
• Classifies DeviceLocked as non-retryable: no more close/reopen churn on a locked device.
• Makes a single try_to_unlock=true handshake attempt so the device itself prompts the user to unlock, then surfaces the result to the caller.

This PR consumes that release and threads the typed signal through to apps.

Changes

• Bump trezor-connect-rs from 0.3.3 to 0.3.4 (both target-gated dependency entries) and refresh Cargo.lock.
• Map the new ThpError::DeviceLocked to the existing typed TrezorError::DeviceBusy in src/modules/trezor/errors.rs, so a locked device surfaces as the back-off signal apps already understand rather than a generic ConnectionError.
• Add a unit test covering the DeviceLocked -> DeviceBusy mapping.
• Regenerate iOS, Android, and Python bindings via ./build.sh all.

Notes

• The UniFFI interface is unchanged: TrezorError::DeviceBusy already existed (Swift case DeviceBusy / Kotlin TrezorException.DeviceBusy, discriminant 4), so only the compiled binding artifacts were rebuilt. No hand-written binding changes.
• Package.swift only has its xcframework.zip checksum updated. No package version bump was performed (./build.sh all without -r); the release/tag bump can happen separately.
• TrezorFeatures.unlocked and trezor_refresh_features() already exist for callers that want to poll lock state.

Testing

• cargo test --lib trezor::errors passes against the published trezor-connect-rs 0.3.4.
• cargo fmt --check clean.
• ./build.sh all completed successfully for iOS, Android, and Python.

Follow-up

The app-side back-off / unlock-prompt UX stays owned by bitkit-ios #613 and bitkit-android #1030, which consume this bitkit-core once released.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 986000b65b

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[coreyphillips]

Release created for testing as v0.3.9 here.

[jvsena42]

@piotr-iohk

[jvsena42]

I'll implement and test on both platforms

[jvsena42]

✅ Validated end-to-end from bitkit-ios #614 (Connect Hardware flow) on a physical Trezor Safe 7, regtest.

• Consumed via bitkit-core v0.3.9, which ships this PR's ThpError::DeviceLocked → TrezorError::DeviceBusy mapping (backed by trezor-connect-rs 0.3.4).
• bitkit-ios now surfaces the typed TrezorError.DeviceBusy (and DeviceLocked) as a dedicated "Trezor is busy — unlock and try again" message instead of a generic connection error.
• QA retest passed the connect flow end-to-end: stale iOS bond shows an actionable error, connect → pairing code → Paired → Finish works, and re-pair after Remove works. No more locked-device close/reopen churn observed.

Approving from the bitkit-ios side

[piotr-iohk]

Same from bitkit-android: synonymdev/bitkit-android#1062