Optimize vcpkg cache with submodule-aware cache key

[magnesj]

Background

Several build jobs have failed recently due to rate limit related to GitHub cache. This is probably related to the large number of cached files by the vcpkg-cache action. This action is now optimized to store a single file for each configuration.

Summary

• Switch vcpkg cache action from TAServers/vcpkg-cache@v3 to CeetronSolutions/vcpkg-cache@copilot/optimize-cache-storage-structure in ResInsightWithCache.yml, clang-tidy.yml, and rhel8-unit-tests.yml.
• Replace token input with a composite cache-key built from runner.os, the compiler, the vcpkg submodule SHA, and a hash of vcpkg.json/vcpkg-configuration.json.
• Add a step that runs git rev-parse HEAD:ThirdParty/vcpkg to capture the submodule commit so the cache is invalidated when the vcpkg submodule pointer is updated.