Add Continuous Fuzzing GitHub Actions workflow

[jverdicc]

Motivation

• Provide scheduled and on-demand continuous fuzzing for core fuzz targets to catch regressions and build a growing corpus of inputs.
• Collect crash artifacts and coverage data per-target so fuzzing results are reproducible and improvement in coverage can be observed over time.

Description

• Add a new workflow at ./github/workflows/continuous-fuzzing.yml that runs on a nightly cron, on pull requests to main, and via workflow_dispatch.
• Configure a matrix of fuzz targets (fuzz_aspec_verify, fuzz_ledger_ops, fuzz_etl_ops, fuzz_etl_read_entry, fuzz_probe_detector) and set fail-fast: false so all targets run even if one fails.
• Install the nightly toolchain plus cargo-fuzz and cargo-llvm-cov, restore and persist per-target corpora with actions/cache, ensure corpus directories, and run cargo fuzz run <target> fuzz/corpus/<target> -- -max_total_time=300 -print_final_stats=1 for each target.
• Upload crash inputs to artifacts on failure and generate per-target coverage via cargo llvm-cov (coverage step marked continue-on-error: true) with coverage reports uploaded as artifacts.

Testing

• Ran cargo fmt --check which succeeded.
• Ran cargo clippy --workspace --all-targets -- -D warnings which succeeded.
• Started cargo test --workspace; the test/build process began but did not complete within this environment before it was manually terminated.

---

Codex Task