docs: document wallet chain-read control methods

[MichaelTaylor3d]

What changed

Closes the documentation gap in #2392: dig-node's control plane now serves five control.wallet.* methods, but only control.wallet.balance was documented.

• docs/run-a-node/manage.md — restructured the single control.wallet.balance section into a control.wallet.* overview (open-read vs token-gated split, the "never signs" invariant) followed by five subsections: balance (unchanged content), coins, peak, coinById, broadcast. Each documents its request/result shape, field table, and the load-bearing semantics from the spec (coins: [] vs an unreachable-chain error, peak_height: null as unknown-not-zero, coinById's ability to see a spent coin, broadcast's accepted-vs-confirmed distinction).
• docs/support/error-codes.md — generalized the -32040..-32043 row descriptions and links from "wallet-balance read codes" to cover all control.wallet.* methods; no codes added or removed.
• scripts/dig-spec.mjs — generalized the matching meaning strings feeding static/error-codes.json so the machine-readable catalog and the prose stay in sync.
• static/error-codes.json — regenerated (npm run machine-specs); the drift gate passed (65 codes, 4 surfaces — same counts as before, wording only).
• package.json — version bump 0.15.0 → 0.16.0 (minor: new documented capability).

Saturation search

Searched the whole docs tree for control.wallet / wallet.balance / control.* summaries. No other page lists or summarizes individual control methods — docs/rpc/methods.md's dig.methods example is the network-profile method list and doesn't include control.* (admin-plane, not network profile), so it did not need updating. docs/run-a-node/manage.md's own control.* bullet summary (status/cache/peers) now has a fourth bullet for wallet.

Verified

• node scripts/gen-machine-specs.mjs — drift gate passed, no code-set changes.
• npm run build — full Docusaurus build (all 15 locales) completed with exit 0, no broken-link errors introduced.

Blast radius

Docs-only change, one repo (docs.dig.net), no code/API touched. The only "existing claim" I checked for mismatch against the given method shapes: the prior control.wallet.balance doc's error table and source/synced/peak_height semantics — carried forward unchanged, no discrepancy found.

Co-Authored-By: Claude noreply@anthropic.com

[MichaelTaylor3d]

HOLD — do not merge until control.wallet.coinById is actually served

This PR documents five control.wallet.* methods. Four are live today. control.wallet.coinById is
not — it is still in flight (dig-node-control-interface#7 → 0.7.0, and dig-node#201 to serve it).

docs.dig.net deploys on merge, so merging now would publish documentation for a capability that does not
exist. Documentation is a claim about the system; a documented method that returns -32601 is the same
class of defect as a UI control that looks live and does nothing.

Release order for this change: dig-node-control-interface 0.7.0 publishes → dig-node serves it and
releases → then this merges. The other four methods' docs (balance, coins, peak, broadcast)
are correct today and are only waiting on the fifth for this reason — they shipped in dig-node#200 with
no docs at all, which is the gap this PR also closes.

Held by the #2392 lane orchestrator, which will merge it once the serving release is green and live.
Refs DIG-Network/dig_ecosystem#2392.

[MichaelTaylor3d]

Converted to draft deliberately — this is a HOLD, not a problem with the PR.

The content is clean and the checks are green. It is blocked on a sequencing fact, not a defect: it documents control.wallet.coinById, which dig-node does not serve yet. Publishing user-facing docs for an unserved method would make docs.dig.net assert something false about the running system.

Merge order:

1. dig-node-control-interface Auto-redirect first-time visitors to their browser locale #7 -> 0.7.0 on crates.io (declares the method) — in flight now.
2. dig-node#201 adopts 0.7.0 and actually serves it.
3. This PR, once a real node answers the method it describes.

It was sitting CLEAN and non-draft, so the hold was purely a note somebody had to read and honour — the same conventional-not-structural shape the review gate flagged on #7 itself. Draft makes it unmergeable until step 2 lands. Mark ready then; no rebase or content change is expected.

[MichaelTaylor3d]

Merge-order constraint — this PR must land AFTER dig-node serves control.wallet.coinById, never before.

This PR documents control.wallet.coinById (10 references in the diff). The method does not exist on any released dig-node: it is implemented on DIG-Network/dig-node PR #201 (branch feat/2392-coin-read-by-id, dig_ecosystem#2392) and the contract half is dig-node-control-interface PR #7 → 0.7.0, neither merged yet.

Merging this first would publish user-facing documentation for a method that answers -32601 METHOD_NOT_FOUND on every node in the field. Per §4.3, docs are authoritative for users — a documented method that does not exist is worse than an undocumented one that does.

Order: contract 0.7.0 publishes → dig-node #201 merges and releases → this PR merges.

The PR is CLEAN and green; nothing here needs changing. It is purely a sequencing hold. Posted from the dig-node lane so the constraint survives outside a session's context.