Docker image layer caching

[gloskull]

Motivation
Reduce CI build time by reusing dependency layers across the repository's Rust, contracts, WASM, and JS services.
Provide an explicit cache-warming job that validates buildability and populates BuildKit/GHA caches without publishing images.
Keep Docker build contexts small and safer by excluding local deps, build outputs, and VCS metadata.
Description
Add Dockerfile.ci with BuildKit-enabled, multi-stage targets and cache mounts for Cargo registries, Cargo git checkouts, Cargo target directories, and npm caches, covering rust-root-test, contracts-test, wasm-build, meter-simulator-test, and usage-dashboard-build targets.
Add a GitHub Actions docker-layer-cache matrix job that runs docker/build-push-action@v6 with cache-from/cache-to set to type=gha and a per-target scope to persist and restore layer metadata.
Add .dockerignore to exclude .git, .github, **/target, **/node_modules, .next, coverage and other local/build artifacts from Docker contexts.
Add docs/CI_DOCKER_LAYER_CACHING.md runbook documenting architecture, CI behavior, monitoring, rollout (blue/green + canary) and security notes, and add small placeholders in the Dockerfile to allow cargo fetch for workspace crates.
Use an ARG RUST_VERSION=stable and standardize node image usage for JS targets to ensure reproducible builds.
Testing
Ran cargo fmt --all -- --check and it succeeded.
Ran cargo clippy --all-targets --all-features -- -D warnings and it succeeded.
Ran cargo test --locked and it succeeded.
Verified workflow YAML parsing with a Ruby check and git diff --check; both checks succeeded, and local Docker builds were skipped because the Docker CLI is not available in this environment.
Closes #82