CI: Windows jobs can fail in rust-cache before tests run

[thepastaclaw]

Summary

Windows CI jobs can fail before any Rust build/test command runs because the Swatinem/rust-cache@v2 step exits with failure while only printing cache configuration and ... Restoring cache ....

Observed examples:

• PR feat(dash-spv): devnet LLMQ routing overrides (chainlocks/isd24/platform) #786 Windows / ffi: https://github.com/dashpay/rust-dashcore/actions/runs/26536128815/job/78165364578
• dev branch Windows / rpc: https://github.com/dashpay/rust-dashcore/actions/runs/26528136761/job/78137248833

Evidence

For PR #786:

• failing step: Run Swatinem/rust-cache@v2
• skipped after that: install/test steps, including Run tests
• check annotations only show GitHub runner notices/warnings, no source/test failure
• PR feat(dash-spv): devnet LLMQ routing overrides (chainlocks/isd24/platform) #786 only changes:
  • dash-spv/src/client/config.rs
  • dash-spv/src/main.rs
  • dash/src/sml/llmq_type/mod.rs
  • dash/src/sml/llmq_type/network.rs

For current dev (58d61ea50ac4c217c8e1350b57363b351a0b99d4), the same cache step failed in a different Windows job (Windows / rpc) before build/test work, which suggests this is not caused by PR #786's code changes.

Impact

Affected PRs can remain red even though no build/test command ran. Maintainers with Actions permissions may need to rerun the failed job/run, and we may want to harden the cache step if this repeats.