Bump bittensor stack and drop substrate-interface

[seroperson]

Summary

Bumps bittensor 10.0.1 -> 10.3.1, bittensor-cli 9.17.0 -> 9.21.1, bittensor-wallet 4.0.0 -> 4.0.1, and removes the direct substrate-interface dependency entirely. The new bittensor stack ships async_substrate_interface and the SS58/keypair helpers we were pulling from substrate-interface, so the dep is redundant.

Knock-on refactors:

• gittensor/cli/issue_commands/helpers.py: SS58 validation now uses bittensor_wallet.utils.is_valid_ss58_address (returns bool) instead of substrateinterface.utils.ss58.ss58_decode (raises). Substrate client imports switched to async_substrate_interface.SubstrateInterface
• gittensor/cli/issue_commands/view.py: SubstrateInterface import swap, no behavior change
• gittensor/validator/issue_competitions/contract_client.py: imports Keypair from bittensor_wallet and ExtrinsicNotFound from async_substrate_interface.errors; drops the dual-exception shim. Adds register_issue(...) so the CLI can reuse the same code path as the rest of the contract operations
• gittensor/cli/issue_commands/mutations.py: issues register now drives the contract through the existing IssueCompetitionContractClient instead of constructing ContractInstance from a local ink! artifact. This removes the runtime requirement to have smart-contracts/issues-v0/target/ink/issue_bounty_manager.contract built locally before registering an issue

Tests:

• tests/cli/test_cli_helpers.py: removed the TestCliRegisterLogicalFailures class, whose only test asserted the "Contract metadata not found" branch that no longer exists
• tests/cli/test_cli_json_error_output.py: patch targets updated to async_substrate_interface.SubstrateInterface

Related Issues

Closes #1334, #1338

Type of Change

• Bug fix
• New feature
• Refactor
• Documentation
• Other (dependency bump and accompanying refactor to keep call sites compiling)

Testing

• Tests added/updated
• Manually tested

Basic things were tested, but not the chain interaction itself. The problem is that there are no real e2e tests against a real chain and no any documented easy way to setup a local chain. Unit tests use mocks, so they won't show anything.

There are references to up.sh, which is likely intended to setup a local chain, but it doesn't exist in the repo. So I guess it should be either tested e2e locally by whoever has this up.sh (would be good to share it at least), or we should consider setup a necessary infra for e2e testing in this repository. The second variant would be also a good addition, as any chain-related changes are literally untested.

Checklist

• Code follows project style guidelines
• Self-review completed
• Changes are documented (if applicable)

Note

pyright (pre-push hook) reports 234 reportPrivateImportUsage errors against bittensor.logging, bittensor.Subtensor, bittensor.Wallet, etc. These were 0 on current test branch. The new bittensor 10.3.1 re-exports via from .submodule import X # noqa: F401 without an explicit __all__, which pyright treats as private. Runtime imports still work. I've relaxed reportPrivateImportUsage in pyright config as for now, but actually the fix should relate to bittensor's side.

[xiao-xiao-mao]

You already have 3 open PRs in this repo. The threshold is 3, so closing this one until a maintainer can triage your existing PRs. Please don't open any more — they'll be closed too.

[seroperson]

Great. How would I know this? @anderdc

[anderdc]

fix conflicts

[anderdc]

Two blockers before this can merge.

1. Drop the dead SS58 regex fallback (gittensor/cli/issue_commands/helpers.py:486-512).

With substrate-interface gone, bittensor_wallet is a hard pinned dep (bittensor-wallet==4.0.1) and is imported unconditionally elsewhere — module-level at gittensor/validator/issue_competitions/contract_client.py:15 and inside the register try block at gittensor/cli/issue_commands/mutations.py:178. If bittensor_wallet.utils weren't importable, the whole CLI would already be broken at module load. The except ImportError: pass branch and the regex fallback are unreachable.

Lift from bittensor_wallet.utils import is_valid_ss58_address to module level, drop the try/except + regex fallback. The function collapses to ~10 lines.

2. Restore visibility in the register error path (gittensor/cli/issue_commands/mutations.py:218-222).

client.register_issue(...) returning None collapses three distinct failure modes — contract revert, the insufficient-balance pre-check (free_balance < 100_000_000 in _exec_contract_raw), and any exception during compose/sign/submit — into a single unconditional _print_register_revert_hints() call. The real error string only goes to bt.logging.error(...), and the extrinsic hash that the old code printed on revert is dropped entirely.

An operator hitting a non-revert failure (RPC timeout, fee shortage) sees misleading "Common revert reasons" hints with no way to inspect what actually happened on-chain. For real reverts, the extrinsic was submitted — the hash is how you'd inspect it via a block explorer.

Two ways to fix, your call which:

• Return a richer type from _exec_contract_raw — e.g. Tuple[Optional[str], Optional[str]] for (hash, error_msg) — so the CLI can branch revert-vs-raw-error and always print the hash when one exists.
• Less invasive: in issue_register, set bt.logging to at least INFO before the client.register_issue(...) call so the underlying f'{method_name} failed: {result.error_message}' line is visible to the user, and separately have _exec_contract_raw print the extrinsic hash on failure too.

The pyright reportPrivateImportUsage global loosen is fine to keep — bittensor upstream needs __all__; we'll revert once they ship it. Per-line # pyright: ignore at ~234 access sites would be noisier than the workaround is worth.

[seroperson]

• Dead SS58 fallback removed
• Visibility restored in register error path (using Option A)