fix: keep deployments successful when verification fails

[banteg]

Summary

• make explorer verification best-effort after a successful deployment
• preserve strict behavior for explicit NetworkAPI.publish_contract() calls
• cover both account and contract-container deployment entry points

Root cause

Both deployment paths cached a successfully deployed contract and then called the active explorer synchronously. Any explorer error, including a missing explorer, API rejection, timeout, or plugin failure, propagated from deploy() even though the on-chain deployment had already succeeded.

User impact

deploy(..., publish=True) now returns the deployed contract instance when explorer verification fails and logs the original error instead. Deployment tracking and the existing local-network guard keep their current behavior.

Validation

• 137 passed across tests/functional/test_accounts.py and tests/functional/test_contract_container.py
• focused publish regression tests pass for both deployment APIs
• Ruff lint and formatting checks pass for all changed files
• git diff --check passes

Local pre-commit mypy currently reports seven errors in three unrelated files; the same errors reproduce on a clean upstream/main worktree.