fix: report PAT check as inconclusive on transient GitHub /user failures

[statxc]

Summary

When the GitHub /user endpoint is temporarily unreachable, handle_pat_check was reporting stored PATs as invalid (pat_valid = False), which surfaces in gitt miner check as a hard "PAT invalid" message. That can push miners to rotate a token that's actually fine — the validator just couldn't reach GitHub.

This PR keeps the existing failure path for permanent errors and adds a third state for transient failures: pat_valid = None (inconclusive) with a short "please retry in a few minutes" message. The check handler now uses validate_github_credentials_result() so the transient_failure flag introduced in #932 isn't dropped.

The broadcast handler is intentionally not changed — it already rejects the PAT and short-circuits the same way before and after, so swapping the rejection string there isn't worth the extra surface area.

Related Issues

Closes #1106

Type of Change

• Bug fix
• New feature
• Refactor
• Documentation
• Other (describe below)

Testing

• Tests added/updated
• Manually tested

Added test_transient_identity_lookup_reports_inconclusive — patches get_github_identity to return TRANSIENT_FAILURE end-to-end and asserts pat_valid is None, the inconclusive message, and that the GraphQL test query is not called (proves the early return).

Updated test_valid_pat and test_stored_but_invalid_pat to patch the new symbol (validate_github_credentials_result); behavior assertions are unchanged.

Local run:

uv run pytest tests/validator/test_pat_handler.py -v   # 22 passed
uv run pytest tests/                                   # 1450 passed
uv run ruff check  gittensor/validator/pat_handler.py tests/validator/test_pat_handler.py
uv run ruff format --check gittensor/validator/pat_handler.py tests/validator/test_pat_handler.py
uv run pyright   gittensor/validator/pat_handler.py tests/validator/test_pat_handler.py

Checklist

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

[xiao-xiao-mao]

duplicate of #1107.

[statxc]

@anderdc Hi, could you please take look at this PR carefully?
It handles same issue 1107 but it’s not actually the same change. I think this one is more solid. Could you please reopen it? thanks