Can't let one rpc fail fuck things up.

[br0wnD3v]

3-Layered RPC Fallback for Transaction Sending

Adds backup and public RPC endpoints as fallback layers when the primary RPC fails to send oracle price updates.

Problem

MrOracle uses a single RPC endpoint with no failover. If it goes down, oracle price updates stop silently. A 13-hour RPC
outage meant no prices were pushed on-chain, which blocked all user position operations (close, SL, TP) and led to a
preventable situation.

Solution

New fallback chain for every price update transaction :
Primary RPC → Backup RPC → Public RPC (api.mainnet-beta.solana.com)

Each layer: get blockhash → sign → send → confirm (2 polls @ 500ms) → next layer if failed.

Changes

• New rpc_fallback.rs module with RpcFallback::sign_and_send()
• Two new CLI args: --rpc-backup (optional) and --rpc-public (defaults to mainnet-beta)
• update_pool_aum rewritten to build raw instructions via InstructionData + ToAccountMetas instead of anchor
  RequestBuilder, then delegates to rpc_fallback.sign_and_send()
• Explicit error logging per layer: [Price Update] PRIMARY RPC failed: ...
• Final message on total failure: All RPCs failed. Please handle ASAP. Critical Priority.

What's NOT changed

• Core 5-second loop timing
• Priority fee polling
• Database price fetching (with retry)
• Pool account fetching at startup
• let _ = non-blocking error handling in the main loop

Usage

--rpc-backup "https://adrena-solanam-6f0c.mainnet.rpcpool.com/"

Without --rpc-backup: primary → public. With it: primary → backup → public.

[AamirAlam]

LGTM