fix(server): retry all PD peers while waiting for storage

[bitflicker64]

Kept in sync with hugegraph#175

Purpose of the PR

• close [Bug] wait-storage.sh binds to the first PD answering /v1/health and never fails over to other peers #3123

wait-storage.sh selected the first PD answering /v1/health, then polled
/v1/stores only on that peer. Since the health endpoint can succeed before
the peer has raft-backed store state, a storeless first peer could block Server
startup even when another configured PD already reported an Up store.

Main Changes

• Probe /v1/stores across every configured PD peer on every retry.
• Succeed as soon as any peer reports a store in state Up, without a separate
  /v1/health selection stage.
• Preserve authentication, configured peer order, custom REST endpoints, retry
  timing, and the existing outer timeout failure.
• Add a deterministic shell regression suite and an isolated Server CI job.

Verifying these changes

• Need tests and can be verified as follows:
  • test-wait-storage.sh: 4 scenarios passed, 0 failed.
  • The same suite fails against the unmodified script on the pinned first peer.
  • bash -n passed for the production and test scripts.
  • Server CI workflow YAML parsing and git diff --check passed.
  • Apache RAT and EditorConfig checks passed for the 20-module dependency set.

Does this PR potentially affect the following parts?

• Dependencies
• Modify configurations
• The public API
• Other affects (HStore Server startup readiness)
• Nope

Documentation Status

• Doc - TODO
• Doc - Done
• Doc - No Need

Out of scope

• PD and Store health endpoint semantics
• wait-partition.sh
• Environment-variable and timeout configurability cleanup
• Docker Compose health checks and documentation