lab-ops-hub: container validator still allows IDNA-disallowed Unicode hosts (space separators, bidi/format controls, U+FFFD) the Worker validator refuses #516

Summary

While verifying the fix for #510, an extended character probe (beyond #510's battery of printable ASCII + 0x7F–0xA0 + é + a combining char) found a further divergence class between the two lab-ops-hub crawl validators: hosts containing certain Unicode characters are refused by the Worker's validateCrawlUrl (pkgs/lab-ops-hub/src/lib.ts — the WHATWG URL parser's IDNA processing rejects them) but are still certified as "a public DNS name" by the container's validate_crawl_url (pkgs/lab-ops-hub/container/app.py).

Verified on main 025794c8 (plus the #510 fix, which does not affect these codepoints) by driving both real validators with each character embedded as http://exam<char>ple.com/:

Controls, identical on both sides (must not change): U+00AD (soft hyphen), U+180E, U+200B (ZWSP), U+2060, U+2064, U+FEFF are ALLOW/ALLOW; ordinary DNS names, exam_ple.com, valid IDN hosts (e.g. examéple.com) and valid IPv6 literals are ALLOW/ALLOW.

Why it matters

Same parity premise as #468/#502/#505/#507/#510: the container validator is the defence-in-depth re-check on the component that actually fetches (#441), so where the two validators disagree the effective policy is the weaker one. These hosts are not valid DNS names under UTS-46/IDNA either — no resolver accepts a host with a bidi override or an ideographic space in a label — so the container's verdict is meaningless for them and the crawl fails below the validator instead of at it.

This is a wider, IDNA-shaped class than the ASCII/codepoint-range fixes so far: a correct fix probably needs to judge the host label-by-label against the IDNA disallowed set (or attempt a strict UTS-46/IDNA encoding of each label and refuse on failure), not another codepoint range. Python's stdlib encodings.idna implements IDNA 2003, not UTS-46, so its verdicts need comparing against the WHATWG set before relying on it.

Acceptance criteria

• validate_crawl_url refuses a host containing any of the codepoints in the table above, up front with a clear validator error.
• Controls unchanged: the ALLOW/ALLOW codepoints in the table, ordinary DNS names, exam_ple.com, valid IDN hosts, valid IPv6 literals and public targets are still accepted on both sides.
• A regression test in pkgs/lab-ops-hub/test/test_container_validate.py covers each refused class, and fails with only container/app.py reverted (ablation).
• The Python and TS validators agree on the extended battery above (the table's codepoints plus the lab-ops-hub: container validator still allows DEL/C1-control, NBSP and WHATWG-forbidden (<>^|) hosts the Worker validator refuses #510 battery).

Cross-repo note: this validator pair lives only in developing-today/code (see #510's cross-repo check), so this is a single-repo issue.