Sync to kth/0.80.0 + full v2 regeneration of py-native bindings

[fpelliccioni]

Summary

Moves py-native off kth/0.79.0 and regenerates every Python wrapper through the v2 generator against the current C-API surface. Three layers in one PR because they cannot land separately without leaving the tree in a non-building intermediate state (bumping conan alone breaks the build, regenerating alone doesn't relink module.c, fixing chain.cpp alone doesn't compile against the renamed types…).

1. Conan + build plumbing

• conanfile.py → kth/0.80.0 (was 0.79.0).
• setup.py sources list now reflects the full generated surface (binary + new chain classes + new wallet classes). src/chain/history.cpp is out of sources — the generator covers that surface via history_compact + history_compact_list. chain.cpp stays (async-callback bridge to safe_chain, doesn't fit ClassConfig).
• scripts/build-nix.sh — new NixOS-only wrapper: resolves gcc.cc.lib into LD_LIBRARY_PATH, provisions a local .venv so editable installs have a writable prefix, scrubs stale build/ / kth/ / conan artefacts before every run, and runs the pytest suite at the end (skippable with --no-tests). Non-NixOS devs keep using scripts/build.sh unchanged.

2. Regenerated surface (~70 files)

Full generate_capi_v2.py --all --target python --write:

• Modified every existing binding to pick up the C-API changes from kth 0.79 → 0.80: handle typedefs post the primitives.h split (kth_*_mut_t / _const_t), the *_simple convention on flag-aware overloads, the leak / leak_if_valid adjustment on types without operator bool, and the owned-copy policy for scalar-value list accessors (short_ids, mnemonics).
• Created ~38 new bindings that weren't generated before: history_compact{,_list}, double_spend_proof{,_spender}, get_blocks, get_headers, operation{,_list}, prefilled_transaction{,_list}, token_data, utxo{,_list} on the chain side, and ec_private, ec_public, hd_private, hd_public, payment_address_list, wallet_data on the wallet side.
• kth_native.pyi and src/module.c picked up the splice updates for every class above.

3. Hand-fixes in the surviving hand-written wrappers

Where the regen doesn't reach and the C-API renames broke the build:

• src/chain/chain.cpp — kth_merkleblock_t → kth_merkle_block_mut_t, and the matching _mut_t adjustments on history_compact_list, stealth_compact_list, compact_block, payment_address; kth_wallet_payment_address_construct_from_string → _construct_from_address.
• src/chain/history.cpp — list / element handle renames to *_mut_t. The file is kept for its async surface but is out of setup.py sources; the generator covers the sync side.
• src/module.c — dropped 50 lines of stale PyMethodDef entries (for merkle_block, history_compact{,_list}, compact_block, stealth_compact{,_list}, payment_address) that pointed at pre-regen function names. Every class on that list is now registered through the AUTO-generated PyModule_AddFunctions block further down.

Test plan

• scripts/build-nix.sh builds the .so cleanly on NixOS.
• Existing pytest suite stays green: 89 passed in 1.23s.
• CI wheel builds on the standard linux / macOS matrix.

Follow-up

End-to-end pytest cases for the ~30 newly-generator-driven classes (double_spend_proof, get_blocks, get_headers, operation, prefilled_transaction, token_data, utxo, history_compact, ec_private, ec_public, hd_private, hd_public, wallet_data, and the matching _list types). They compile, link and import today; dedicated Python-side tests will land in a separate PR.

Summary by CodeRabbit

• New Features

  • Much-expanded Python API: new chain types (compact/merkle/get-blocks/headers, operations, token data, UTXO, prefilled txs, history, double-spend proofs) and wallet types (EC/HD keys, payment addresses, wallet data), plus script helpers.

• Improvements

  • Core dependency bumped to 0.80.0.
  • Safer resource lifecycles, better input validation, and clearer error reporting.
  • Added a Nix-friendly build wrapper for reproducible builds.

• Bug Fixes

  • Hardened bindings to prevent crashes and reduce allocation/edge-case failures.

• Tests

  • Large suite of new tests validating lifetimes, bindings, serialization, and behavior.