Add latest ByteDance CodeSignal OA questions

[tomtoto757]

Summary

• add four verified ByteDance CodeSignal OA questions from the Aug 25, 2026 sighting
• preserve each canonical FastPrep title, including the requested (for mle also :) suffix
• re-sort the complete question bank and refresh date badges with the repository freshness script
• keep the Stage 4 change scoped to README.md

Verification

• production FastPrep catalog readback matched all four canonical IDs, titles, ByteDance attribution, Aug 25 latest sighting, and algorithm format
• all four exact public FastPrep routes returned HTTP 200
• python3 scripts/refresh-freshness.py — second pass was idempotent with 0 marker changes and 0 reordered rows
• PYTHONDONTWRITEBYTECODE=1 python3 -m unittest -v scripts/test_sync_practice_formats.py — 10/10 passed
• README audit — 1,990 canonical rows and 1,990 unique matching FastPrep routes; no malformed rows, duplicate routes, date-order violations, company-list duplicates, or company-list order violations
• target audit — exactly one ByteDance Coding row per canonical route, with the exact catalog title, two matching URLs, six pipe characters, and 🔥 Aug 25, 2026
• normalized diff audit — four intended target additions, 82 pre-existing rows changed only by deterministic freshness-marker aging, no removed routes, and no other content changes
• git diff --check — passed

Concurrent README ownership

Open draft PRs #147–#150 also touch README.md, but none contains or owns any of these four ByteDance routes. This focused branch starts from the latest master (3c98ad57c68670b3cfb001a50c7818d470548683) and intentionally does not absorb their unrelated IBM, Anthropic, Walmart, or Amazon changes.

Independent regression review

An independent read-only reviewer checked exact base 3c98ad57c68670b3cfb001a50c7818d470548683 and exact head d0344d2ad2a78676268987da7601d7df01a0ce18; no PR-introduced findings.

• confirmed the four exact catalog-backed ByteDance rows, distinct company-route identity for the two titles also represented under TikTok, and HTTP 200 rendered routes
• confirmed 1,990 rows / 1,990 unique routes with no malformed rows, duplicate routes, bad link pairs, date-order violations, or same-date reordering
• independently classified the other 82 replacements as exactly 15 🔥 → 🆕 and 67 🆕 → unbadged freshness changes
• reran freshness idempotence twice, the 10 format-sync unit tests, and git diff --check; all passed and the reviewed head remained byte-stable
• verified remote master and the PR branch still equal the reviewed SHAs, and PR Add latest ByteDance CodeSignal OA questions #151 is draft, clean, and mergeable

Draft status and remaining risk

python3 scripts/sync-practice-formats.py --check reports four stale generated format pages on both the exact base and this head. Both runs report the same four pages, 148 catalog-missing routes, and zero unclassified rows. The generated README Coding label remains 1,772 while the exact base contains 1,777 Coding rows and this head contains 1,781, an increase of exactly the four intended additions. The inherited company summary likewise remains unchanged at 214+ / +204 while both SHAs contain the same 215 unique sorted company names; ByteDance already occurs exactly once.

Those counters are owned by scripts/sync-practice-formats.py, which regenerates the README counter and all formats/*.md pages together. Manually changing only the README counter would leave the generated Coding page inconsistent, while running the full generator would absorb unrelated live-catalog drift beyond this focused README-only Stage 4 scope. The required row freshness validator passes every row and is idempotent; the generated-count drift is therefore documented rather than partially rewritten here.

The PR remains draft under the Stage 4 default pending maintainer review.

Expected runtime impact: none; this changes README catalog metadata and deterministic freshness markers only.

[tomtoto757]

Superseded by consolidated batch PR #150 at exact reviewed head 3a31217. PR #150 preserves all four ByteDance rows byte-for-byte alongside the three Amazon updates; independent final-SHA review confirmed no content loss. #150 remains draft only for the inherited generated-format synchronization debt.