chore(bench): replace WARP with WAGO runtime

[JairusSW]

Summary

• replace the cross-runtime benchmark's WARP integration with WAGO
• add a small Go host using WAGO's public API and enforced guard-page bounds mode
• generate WAGO benchmark modules with embedded classic payloads
• update runtime result collection, chart labels, and documentation
• remove the old WARP C++ host and generator

Why

The runtime comparison should cover WAGO instead of WARP while preserving guest-side timing and the existing bench() result format. Building the host with wago_guardpage is required for practical performance on the allocation-heavy classic workloads.

Impact

npm run bench:runtimes now auto-builds the WAGO host from WAGO_SRC, writes results under build/logs/runtimes/wago, and includes WAGO in the runtime charts. The separate warpo compiler comparison is unchanged.

Validation

• ran the complete Twitter, CITM, and Canada matrix across WAGO, Wasmtime, WAVM, wazero, V8, and Bun; every runtime emitted three results per payload
• ran npm run charts:runtimes and inspected the regenerated WAGO chart
• built the WAGO host with -tags wago_guardpage
• passed repository pre-commit verification: transform build, Prettier, and ESLint
• passed shell syntax, Node syntax, JSON result, formatting, and diff checks