docs: document official chain IDs and canonical testnet explorer

[memosr]

Adds a "Networks" section to the README with Arc's official chain IDs and a
wallet connection table for Testnet.

Two corrections this fixes:

1. Chain ID. Arc Testnet is 5042002 (0x4cef52). Several third-party chain
   registries and community guides list 1516, which causes wallet connection
   failures. The section includes an eth_chainId curl call so readers can
   verify against a live node.

2. Explorer URL. The canonical testnet explorer is https://testnet.arcscan.app.
   Older guides point at explorer.testnet.arc.network (unreachable) and
   explorer.arc.io (login-gated), neither of which is publicly usable.

README only, no code changes.

Split out of #195 so the security fix there can be reviewed on its own.

[osr21]

Verified every factual claim in this PR independently before reviewing:

Chain IDs — all three match the source of truth. crates/shared/src/chain_ids.rs declares MAINNET_CHAIN_ID = 5042, DEVNET_CHAIN_ID = 5042001, TESTNET_CHAIN_ID = 5042002, and the hex conversions in your table are correct (0x13b2 / 0x4cef51 / 0x4cef52). Live check just now: eth_chainId against https://rpc.testnet.arc.network returns 0x4cef52, so the curl verification block in the callout produces exactly the output it promises.

Explorer claims — confirmed. testnet.arcscan.app responds 200; explorer.testnet.arc.network doesn't resolve at all (NXDOMAIN, so "unreachable" is if anything understated — it's gone from DNS); explorer.arc.io 302s behind Cloudflare, consistent with the login gate you describe. The warning against putting either in blockExplorerUrls is well-founded.

One correction to the framing of the 1516 claim, worth fixing in the callout text. I checked the primary registry (chainid.network / chainlist): 1516 is Story Odyssey Testnet — it has never been an Arc ID there, and the registry's actual Arc Network Testnet entry (5042002 → testnet.arcscan.app) is already correct. What the registry does contain is a likelier source of the confusion: chain IDs 1243/1244 are registered as "ARC Mainnet"/"ARC Testnet" — an unrelated project with a colliding name. So the guides quoting wrong IDs are probably mixing up chains named Arc/ARC, not reading a registry that lists 1516 for this one. I'd soften "several third-party chain registries … list 1516" to something like "some community guides circulate wrong chain IDs (e.g. 1516, which belongs to Story Odyssey Testnet, or 1243/1244, an unrelated project also named ARC)" — that keeps the warning while staying accurate about the canonical registry, which gets this chain right and is worth pointing readers toward rather than away from.

Placement notes:

• No collision with #260 (my mainnet-prep docs PR): that adds chain IDs to docs/running-an-arc-node.md / installation.md for operators choosing a chainspec; this puts them in the README for people connecting wallets. Complementary — though whichever merges second might add a one-line cross-reference so the two tables can't silently drift.
• Fair warning on scope: maintainers here have consistently routed developer-facing content (dapp guides, wallet flows) to developers.circle.com and kept this repo's docs operator-focused. The chain-ID table itself is squarely operator-relevant (it's what --chain selects), but the wallet-connection table is the kind of content that has previously been redirected. I'd keep the PR as-is and let maintainers make that call, but don't be surprised if that half gets trimmed.

Also confirmed the split from #195 is clean — that PR now carries only the Fin-sequence saturation fix, which deserves the isolated review. Nice discipline separating them.

[memosr]

Community guides sometimes circulate wrong chain IDs for Arc, usually by
confusing it with similarly named chains: 1516 is Story Odyssey Testnet, and
1243/1244 belong to a deprecated, unrelated project also called ARC. The
canonical registry (chainid.network) already lists Arc Testnet correctly. This
section gives readers the right values plus an eth_chainId call to verify
against a live node.

[osr21]

Re-verified d6dab01 — the revised callout is now accurate on every axis I checked:

• The registry attribution is fixed and inverted into a strength: instead of impugning chainid.network, the callout now cites it as corroboration ("already lists this correctly"), which gives readers two independent verification paths — the canonical registry and the live eth_chainId call. That's strictly better than the original framing even apart from accuracy.
• The misattribution sources are now named precisely: 1516 → Story Odyssey Testnet, 1243/1244 → the unrelated deprecated ARC project. Naming the actual owners of the wrong IDs is what makes the warning durable — a reader who finds 1516 in some guide can now confirm for themselves it belongs to a different chain, rather than taking the README's word for it.
• Everything else I verified on the previous head is unchanged and still correct: the three chain IDs match crates/shared/src/chain_ids.rs, the hex conversions hold, the live RPC returns 0x4cef52, arcscan responds, and the two dead explorer URLs remain dead.

Nothing left from my review. The remaining open considerations are both maintainer calls, not correctness issues: whether the wallet-connection half stays in this repo or gets routed to developers.circle.com per the usual docs-scope convention, and a one-line cross-reference between this table and the operator-facing one in #260 (whichever merges second). Content-wise this is ready.