telemetry: submit BGP session status onchain per user

[juan-malbeclabs]

Resolves: #3465

Summary of Changes

• Adds a BGP status submitter to the telemetry daemon that periodically checks BGP session state in the BGP VRF namespace and submits `Up`/`Down` status onchain for each activated user on the device
• Adds `BGPStatus` type and `SetUserBGPStatus` executor instruction to the Go serviceability SDK
• Fixes a gap where a disappeared tunnel interface (e.g., after an ungraceful daemon kill) would silently skip the `Down` submission — now correctly transitions from `Up` to `Down` when the tunnel is gone but the user remains activated onchain
• Introduces a `CachingFetcher` in the telemetry daemon that deduplicates `GetProgramData` RPC calls across the BGP status submitter, ledger peer discovery, and collector within a 5s TTL window using `singleflight`, mirroring the pattern in the client daemon
• Adds Prometheus metrics for both the caching layer and onchain submission outcomes
• Adds an e2e test (`TestE2E_UserBGPStatus`) verifying the full `Up` → `Down` lifecycle

Diff Breakdown

Category	Files	Lines (+/-)	Net
Core logic	4	+526 / -0	+526
Scaffolding	5	+161 / -7	+154
Tests	4	+770 / -0	+770
Docs	1	+4 / -0	+4

Mostly new code: a self-contained submitter package, a shared caching layer, and their tests.

Key files (click to expand)

• `controlplane/telemetry/internal/bgpstatus/bgpstatus.go` — `Submitter` struct, `Config`, and pure helpers (`buildEstablishedIPSet`, `computeEffectiveStatus`, `shouldSubmit`)
• `controlplane/telemetry/internal/bgpstatus/submitter.go` — tick/worker loop: collects BGP socket state, resolves tunnel IPs per user, enqueues status updates with retry; includes tunnel-not-found → Down fix
• `e2e/user_bgp_status_test.go` — e2e test validating `Up` detection after BGP session establishes and `Down` detection after `pkill -9 doublezerod`
• `controlplane/telemetry/internal/bgpstatus/submitter_test.go` — unit tests for tick logic, grace period, periodic refresh, in-flight deduplication, and tunnel-not-found → Down transition
• `controlplane/telemetry/internal/serviceability/cache.go` — `CachingFetcher`: `sync.RWMutex` fast-path + `singleflight` slow-path + stale-on-error fallback for `GetProgramData`
• `controlplane/telemetry/cmd/telemetry/main.go` — wires `CachingFetcher` and BGP status submitter; flags: `--bgp-status-enable`, `--bgp-status-interval`, `--bgp-status-refresh-interval`, `--bgp-status-down-grace-period`
• `smartcontract/sdk/go/serviceability/executor.go` — adds `SetUserBGPStatus` executor (instruction opcode 106)
• `smartcontract/sdk/go/serviceability/state.go` — adds `BGPStatus` type with `Unknown/Up/Down` constants and JSON serialization

Testing Verification

• Unit tests cover all tick-loop branches: session established, session down with and without grace period, tunnel not found with last status Up/Down, periodic refresh, in-flight dedup, and submission retry
• `TestE2E_UserBGPStatus` verified end-to-end: connects a client, waits for BGP session to reach `Established` in vrf1, confirms `BGPStatusUp` appears onchain, then kills doublezerod ungracefully and confirms `BGPStatusDown` appears onchain within 60s