fix: only run for prod chains and raise on errors

[gosuto-inzasheru]

last run failed to retrieve pools and gauges for ethereum and gnosis, but the error was caught and the ci workflow continued (https://github.com/BalancerMaxis/bal_addresses/actions/runs/10252037228/job/28361482279)

mainnet
Error using core subgraph for mainnet, skipping: 429 Client Error: Too Many Requests for url: https://api.studio.thegraph.com/query/75376/balancer-v2/version/latest
polygon
arbitrum
optimism
gnosis
Error using core subgraph for gnosis, skipping: 504 Server Error: Gateway Timeout for url: https://api.studio.thegraph.com/query/75376/balancer-gnosis-chain-v2/version/latest
zkevm
goerli
Error using core subgraph for goerli, skipping: 410 Client Error: Gone for url: https://api.thegraph.com/subgraphs/name/balancer-labs/balancer-goerli-v2
Error using gauge subgraph for goerli, skipping: 410 Client Error: Gone for url: https://api.thegraph.com/subgraphs/name/balancer-labs/balancer-gauges-goerli
sepolia
Error using gauge subgraph for sepolia, skipping: Invalid URL 'None': No scheme supplied. Perhaps you meant https://None?
avalanche
fantom
base
mode
fraxtal

this results in a pr which removes all pools/gauges it couldnt retrieve, which i think should not occur ever (#400)

furthermore, the subgraphs for the testnets are not deployed and/or unreliable

suggesting here to only run pools and gauges for production chains, and raise (thus fail the workflow) if data for these chains cannot be retrieved successfully

[Tritium-VLK]

https://github.com/BalancerMaxis/bal_addresses/actions/runs/10266632325

[Tritium-VLK]

https://github.com/BalancerMaxis/bal_addresses/pull/403/files#diff-431ee7bca52fc4815d9d94ad743303c2cf7ddd4222ed29f76826a0d5aae748d2

@gosuto-inzasheru wasn't this supposed to fail instead of removing all the gauges?

[Tritium-VLK]

Only OP is removed, as it is not an chain managed by the Maxis. Gos cleared this up. LGTM.