Re-point test_should_consolidate_to_block_15k at DogeOS chikyu infrastructure

[vmarkushin]

Split out of #38 (defect 3). Since the Tsuki migration (#5) the sync workflow's test uses DOGEOS_CHIKYU but keeps Scroll-Sepolia-era inputs, so it fails at NodeConfig::from_chainspec (crates/primitives/src/node/config.rs:123, Invalid or missing 'startL1Block') before any networking — with or without ALCHEMY_KEY:

• The chikyu genesis (dogeos-reth crates/dogeos-chainspec/res/genesis/chikyu_dogeos.json at the pinned rev 8f0b98b) has no scroll.l1Config.startL1Block and no l1Config.systemContractAddress; chain id 0x5fdaf3 is not a NamedChain, so the extra-fields parser is mandatory and errors out.
• chikyu's l1ChainId is 111111 — the test's https://eth-sepolia.g.alchemy.com URL and scroll-sepolia-blob-data S3 bucket cannot serve it.

Needed to make the sync workflow meaningful again:

1. Add startL1Block and systemContractAddress to the chikyu genesis l1Config in dogeos-reth (and bump the pin here).
2. Choose the L1 RPC endpoint and blob source for chikyu's L1 (chain 111111) and thread them through the test config (ALCHEMY_KEY becomes a generic L1_RPC_URL-style secret, or is retired).
3. Record a new golden block hash at a chosen consolidation height and re-enable the sync workflow's push trigger (gated to workflow_dispatch since Stabilize integration CI startup/event races and fix sync preflight #38).

Until then the workflow is dispatch-only with a preflight that fails fast on missing/empty configuration (see #38).