fix(gecko): replace dead WBNB/BUSD pool with active USDT/WBNB V3 on BNB

[Flotapponnier]

Symptom

Alertmanager fires geckoterminal head_lag stale on bnb every few hours across all 3 regions in staging and prod:

:warning: geckoterminal head_lag stale on bnb (sgp)
geckoterminal hasn't pushed a head_lag sample for bnb / sgp in over 5 minutes.

Root cause

The BNB pool configured across 3 harnesses uses PoolID: "24" which points to WBNB/BUSD PancakeSwap V2 — near-idle since Binance stopped issuing new BUSD in September 2024.

Staging-sgp log evidence (last 5000 lines):

• gecko/bnb: 2 samples
• gecko/base: 7 samples
• gecko/solana: 1 sample

The 5-min sample-freshness rule in alert_rules.yml fires because the pool physically has fewer than 1 swap per 5 min.

Fix

Replace with PancakeSwap V3 USDT/WBNB 0.01% — currently the #1 BSC pool by 24h volume ($130M+).

• Pool address: 0x172fcd41e0913e95784454622d1c3724f546f849
• Internal GT pool_id: 160787671 (via app.geckoterminal.com/api/p1/bsc/pools/<address>)

Scope

Same fix applied to all 3 harnesses that hardcode the same pool list:

• aggregator-head-lag (the one firing the alert)
• aggregator-latency-benchmark
• metadata-coverage

Test plan

• go vet clean on the affected files
• After deploy: staging-sgp logs show [HEAD-LAG][GECKO][*][bnb] samples arriving every 30-120 s instead of every 15-30 min
• After deploy: alertmanager stops firing gecko_head_lag_stale{chain="bnb"}