Complete --submit mode: urlProbe, activation, cookies, status_codes, update-existing

[soxoj]

Problem

maigret --submit <url> currently handles only the easy case: clearweb site, no anti-bot, no JSON probe URL, no token-refresh activation, no cookie-based auth, no existing-entry update. Six TODOs in submit.py mark every place where a manual editing fallback is required. The CONTRIBUTING guide explicitly tells contributors to fall back to manual editing for any non-trivial site — this issue tracks the work to lift that limitation.

Sub-tasks

The TODOs are independent enough to land in separate PRs but share enough context to live under one issue:

1. Cookies (maigret/submit.py:218) — pass --cookies-jar-file through check_features_manually. Easy.
2. status_codes (maigret/submit.py:527) — when manual-features fallback is used, also collect HTTP status codes for the claimed and unclaimed URLs so the resulting entry can pick checkType: status_code if appropriate.
3. Shared HTTP client (maigret/submit.py:195) — replace the bespoke session.get block with SimpleAiohttpChecker so submit-mode benefits from the same SSL handling, error mapping, and retries as the main checker.
4. Filter by errors (maigret/submit.py:265) — the inline Cloudflare-detection regex should be moved into the shared errors.detect machinery (maigret/errors.py) and the dialog should be able to handle other anti-bot vendors uniformly.
5. urlProbe / activation (maigret/submit.py:475–476) — currently --submit aborts when the existing site has a urlProbe or activation block. Detect public JSON APIs alongside the HTML check and offer to populate urlProbe. Activation is a bigger lift (per-provider) — at minimum, document the gap inline.
6. Update-existing (maigret/submit.py:424) — the dialog already detects when the same domain has an existing entry but does not actually merge the new evidence into it. Implement the merge.

Code refs

• maigret/submit.py:195–540
• Parallel checker in maigret/checking.py

[Sushanth012]

I'd like to work on this

[felicabrera]

Would be happy to help here! Is PR #2804 still in progress, or should I tackle one of the other sub-tasks?

[juliosuas]

I’ll take subtask 2: collect the claimed and unclaimed HTTP status codes in the manual-features fallback so --submit can choose checkType: status_code when the evidence supports it.

I checked the open PRs first: #2804 covers blocking-pattern detection and #2807 covers cookies, so this should not overlap. I’ll keep the change focused and include regression tests.