[indexer/verifier] Define canonical FlowPulse observation identity

[FlowmemoryAI]

Objective

Specify how an indexer uniquely identifies an observed FlowPulse log after reading transaction receipts and logs.

Allowed folders

• services/indexer/
• services/verifier/
• docs/ARCHITECTURE.md only for a short cross-link
• docs/DECISIONS/ only if the identity format becomes a durable decision

Forbidden folders

• contracts/
• apps/
• hardware/
• crypto/
• research/
• RPC integrations, production indexer runtime, databases, deployment config

Acceptance criteria

• The spec defines canonical fields such as chain id, contract address, block number/hash, transaction hash, log index, and pulse id.
• It states which fields come from receipts/logs rather than contract execution.
• It describes how duplicate, pending, finalized, and reorged observations should be named conceptually.
• No live chain access or database implementation is added.

Risk level

Medium

Recommended worktree/agent

E:\FlowMemory\flowmemory-indexer / Services Agent

Scope guard

Specification only; do not build a production indexer or verifier service.