dash(dashboard): Best Share + peers + found-blocks parity with LTC dashboard (v0.2.4)

[frstrtr]

DASH dashboard → LTC parity (ships with v0.2.4)

Wires the DASH c2pool dashboard cards that were broken/unpopulated to the real DASH work-source / tracker / coin-state, reusing main_ltc's WebServer / MiningInterface wiring rather than forking a parallel dashboard. All changes are display-only — no share/target/payout/consensus logic is touched (confirmed below).

Base: fresh master @ 9ac8b3dd (#781).

PRIORITY-1 — Best Share card (end-to-end, populated)

The operator wants to see how close DASH solo shares got to net difficulty — solo shares rarely cross net-diff, and on the DASH path they seldom even mint onto the sharechain, so the tracker's verified-share m_on_share_difficulty hook (the only best-share feed) never fired → the 🎯 Best Share card sat empty.

Root fix — record best-share on the stratum vardiff-accept path, where every accepted pseudoshare is seen:

• New IWorkSource::record_best_pseudoshare(...) (default no-op → LTC/BTC/DGB unchanged). The core StratumSession::handle_submit calls it for every accepted pseudoshare, passing the already-computed X11 difficulty + miner + the raw submit fields.
• DASHWorkSource overrides it: recomputes the exact X11 pow-hash (shared dash_reconstruct_pow_hash helper, byte-identical to mining_submit/compute_share_difficulty) and forwards (difficulty, miner, pow_hash) to the dashboard via a callback main_dash binds to MiningInterface::record_share_difficulty.
• record_share_difficulty / BestDifficulty / /best_share now also carry the share hash + timestamp; a compact best_share block is added to /local_stats; the dashboard card renders the record-hash line.

Verified live against a testnet dashd (ubuntu@192.168.86.52, isolated instance, testbed only) — a pseudoshare miner submitted and /best_share populated:

{
  "network_difficulty": 0.008288110320964296,
  "round": {
    "difficulty": 0.0007045984339726283,
    "pct_of_block": 8.501315820934325,
    "miner": "yiXsatJHYQD2u2vFVR1yeWQEUn3vDh7Ley",
    "hash": "0000058b39f949586e72742e3798efb456ba44d8f93686f2add980a9a271e78f",
    "timestamp": 1784597678
  },
  "session": { "difficulty": 0.0007045984, "pct_of_block": 8.50, "hash": "0000058b39f9...", ... },
  "all_time": { "difficulty": 0.0007045984, "pct_of_block": 8.50, "hash": "0000058b39f9...", ... }
}

i.e. the best share got to 8.5% of net difficulty, with the exact hash and the miner that found it — precisely the metric requested.

Other cards wired (audit vs LTC dashboard)

Card	Before (DASH)	After
Best Share 🎯	empty ("-") — no stratum feed; tracker hook never fires solo	round/session/all-time difficulty + pct_of_block + hash + miner, populated from every accepted pseudoshare
Peers / node-status	peers {incoming:0, outgoing:0} always — dash::Node had no get_peer_info_json, set_peer_info_fn unwired	real pool p2p peers (added dash::Node::get_peer_info_json, ltc parity; wired set_peer_info_fn)
Recent blocks / found-block history	DASH block wins never recorded to the dashboard	DASHWorkSource fires a found-block hook after dual-path dispatch; main_dash records via record_found_block

Already correct on master and confirmed populated in the same run (reused LTC data path, no change needed): pool/network hashrate, network difficulty, miner count + per-miner rows (miner_hash_rates), shares/efficiency, block value + payout split (block_value 2.381 / block_value_miner 0.595 via coin_work_fn), attempts-to-block, uptime, sharechain tip/stats.

Not in this PR (documented follow-ups; higher-touch, coin-embedded specific): node-topology per-coin panel (set_node_topology_fn is tied to embedded coin_peer_sources) and the PPLNS current-payouts panel. DASH masternode-payee / DIP4 coinbase display has no analogous LTC slot to reuse.

Consensus-neutrality

Dashboard is read-only. The two touches on sensitive paths are strictly additive display feeds:

• stratum_server.cpp: one record_best_pseudoshare(...) call inserted between existing statements on the already-accepted path — no control-flow, acceptance, target, or payout change.
• work_source.cpp: the found-block hook fires after the existing dual-path dispatch decision (!payee_guard_reject) and only reads a callback.

No model/tool authorship attribution anywhere (branch/commit/PR/comments/HTML).

Build / test

• c2pool-dash builds clean (0 errors).
• Live testbed run against testnet dashd: web dashboard LIVE, /best_share + /local_stats populated as above. No production deploy (hotel deploy is an operator tap).

DRAFT — no merge, no deploy.