feat(collection): multi-item DID-attributed mint + bech32 DID decode

[MichaelTaylor3d]

Summary

Closes DIG-Network/dig_ecosystem#199 (dkackman's 203-item collection blocked) and DIG-Network/dig_ecosystem#198 (bech32 did:chia: decode).

collection mint refused any manifest with more than one item. Root cause: Chia's coin-value conservation is bundle-wide, not per-coin — each item's IntermediateLauncher "prints" 1 mojo (a 0-value intermediate coin's own spend creates a 1-mojo singleton launcher), and that mojo must be donated by another coin in the same bundle. The DID's own update spend conserves its exact value, so it can fund at most one item's worth on its own. A real N>1 mint needs a separate XCH funding coin.

What changed

• digstore_chain::collection::build_collection_mint_funded(_in) — reuses the existing (already N-agnostic) build_collection_mint_in per-item loop and additionally spends a separate XCH coin contributing items.len() mojos, returning any excess as change (never silently burned as network fee). Single-item behavior (build_collection_mint) is untouched.
• CLI (collection mint) — selects a funding coin via the existing scan_and_select_funding for items.len() > 1, calls the funded builder, and removes the old >1 item refusal.
• digstore_chain::did::decode_bech32_did + CLI parse_did_arg — accepts a did:chia:1… bech32m address (Sage/CNI's display form) in addition to 64-hex, on every --did flag (collection mint/show, nft mint --did). parse_launcher_id (used for non-DID ids like --nft) is untouched.
• SPEC.md §11.1/§11.2 — documents the --did bech32/hex acceptance and the N=1-vs-N>1 funding contract.
• docs.dig.net — nft-developers.md + command-reference.md updated (bulk manifest minting, bech32 DID acceptance).
• Superproject DEVELOPMENT_LOG.md — durable realization on bundle-wide coin conservation / the IntermediateLauncher funding mechanic.

Blast radius (gitnexus, before editing)

• build_collection_mint_in (upstream): LOW risk, 2 direct callers (collection.rs mint/run) — both are the CLI file I edited; no other consumers.
• parse_launcher_id (upstream): CRITICAL risk, 4 direct callers (nft mint/transfer, collection mint/show) — left completely unchanged; a new parse_did_arg wrapper was added instead, so that blast radius is unaffected.

Verification

• Simulator N>1 proof (the required proof): build_collection_mint_funded_in_validates_on_simulator — a real N=3 DID-attributed mint, funded by a separate 10-mojo XCH coin, validates under Chia consensus in-process; asserts the 7-mojo change coin lands back at the funder (not burned as fee). build_collection_mint_funded_rejects_underfunded_coin proves the insufficient-funds message.
• bech32 DID decode: round-trip test against a known vector + dkackman's own reported DID string; malformed/wrong-prefix rejection tests, at both the chain-crate and CLI layers.
• Installed-binary integration (§3.5): after cargo install --path crates/digstore-cli --force --locked, ran the installed digstore.exe directly (not just cargo test):
  • collection mint with a 203-item manifest (dkackman's exact real-world size) --dry-run: no longer hits the old "single DID-attributed item" refusal; proceeds through funding-coin selection and reaches the same "does not own DID" failure (exit 4) the single-item path already hits under the offline mock.
  • collection show --did did:chia:1s8j4p... decodes to 81e55083864f29bbffed17234f0120b3ba51c69b69a9d785f9bcfda68ebc5d56 (exit 0).
  • Malformed bech32 --did rejected with a clear message (exit 2).
  • Single-item mint regression-checked unchanged (same exit 4 as before).
• Full test suites green: cargo test -p digstore-chain (209 passed), cargo test -p digstore-cli (all suites green, incl. cli_assets.rs 19/19).
• cargo fmt --all -- --check clean; cargo clippy -p digstore-chain -p digstore-cli --all-targets --all-features -- -D warnings clean.
• Coverage: cargo llvm-cov is blocked locally by a pre-existing Windows/cmake toolchain issue unrelated to this change (a libz-sys build script failure under the llvm-cov instrumented profile); this repo has no CI coverage gate configured yet (a separate, pre-existing gap, not introduced here). Every new function has dedicated tests including error paths.

SemVer

Minor (0.9.1 → 0.10.0) — backwards-compatible new capability (feat:): multi-item collection mint + bech32 DID acceptance, no removed/changed public API, no format/schema break.

Follow-up (flagged, not fixed here — separate lane)

chip35_dl_coin's core/src/collection.rs::build_bulk_mint (the hub NFT Studio's on-chain twin) has the same missing N-launcher funding gap — it has no funding-coin parameter at all. Left untouched here per single-writer (a chip35 lane is active on #189); noting on #199 for follow-up.

Test plan

• cargo test -p digstore-chain — 209 passed
• cargo test -p digstore-cli — all suites green
• cargo fmt --all -- --check — clean
• cargo clippy -p digstore-chain -p digstore-cli --all-targets --all-features -- -D warnings — clean
• Simulator N=3 funded collection mint validates under consensus
• Installed-binary smoke test: 203-item --dry-run, bech32 DID decode, malformed DID rejection, single-item regression