Refactor CI to use Subway as RPC proxy/cache instead of Chopsticks

[rockbmb]

This is related to #513 .

I lifted my work from polkadot-fellows/runtimes#1068, and implemented it here, in the CI/Snapshot Update workflows.

Summary:

• new folder in .github/subway-configs; it contains Subway configs for the RPC proxies that will be created during CI: one for each Polkadot/Kusama chain that has test suites containing *.e2e.* in their filenames.
  • E2E tests are the heaviest in terms of network requests, particularly repeated queries for the same data e.g. block hashes and storage; XCM connectivity tests can be superiorly bounded at 10-100 roundtrips, so a cache is not mandatory in these cases.
• Workflow in .github/workflows/ci.yml refactored the Chopsticks-as-cache mechanism introduced in Use chopsticks as cache to avoid RPC stalling in CI workflow #506/Rework CI to use Chopsticks servers to reduce RPC load #504 to instead use Subway for RPC caching per @xlc's suggestion
  • it uses a matrix strategy with two jobs, one for each network: Polkadot and Kusama
    • each of these jobs with share the Subway servers started for chains in that network among themselves, as their endpoints will have been injected into .env
  • it limits concurrency to 2 test files running simultaneously

[github-actions]

Found configuration errors in Subway config files (incorrect endpoints, duplicate entries) and a logic issue in the GitHub Actions workflows where caching is attempted before the repository is cloned.