Self-contain 2023-bids-dicom figures; annex 12 pointer files in pics/

[yarikoptic]

Closes #30.

• 2023-bids-dicom.html: the BIDS-minder slide's upstream URL 404s (the file was removed from bids-starter-kit in [ENH] Move starter kit to new BIDS website with redirects bids-standard/bids-starter-kit#414); use our pics/BIDS-minder.svg. The two ReproIn figures on "DICOMs in BIDS" were also hotlinked from ReproNim/reproin@master; they are now annexed as pics/dbic-flow.png and pics/dbic-conversions.png.
• pics/BIDS-minder.svg and 11 other pics/ files had been committed as git-annex pointer files (unlocked) instead of symlinks. tools/build_preview.py copies such files verbatim, so the PR preview served the /annex/objects/... text instead of the image. git annex lock turned them into symlinks. The keys are unchanged and all 12 are on datasets.datalad.org.

Before merging, make the two new keys available. The git-annex branch was deliberately not pushed from here, so until then the git-annex check fails for exactly these two files. Pushing only git-annex triggers no PR event, so re-run that check afterwards. The copy step keeps the datasets.datalad.org mirror from losing the images it hotlinks today:

git annex registerurl MD5E-s149431--29e378a46a7294f2f7b040a48136c9bf.png \
  https://raw.githubusercontent.com/ReproNim/reproin/fe5488b59c607ead1384da0e3efcbe57770c34e8/docs/source/images/dbic-flow.png
git annex registerurl MD5E-s293267--d7a4fa9b78b172ed35d73251763a045f.png \
  https://raw.githubusercontent.com/ReproNim/reproin/fe5488b59c607ead1384da0e3efcbe57770c34e8/docs/source/images/dbic-conversions.png
git annex get pics/dbic-flow.png pics/dbic-conversions.png
git annex copy --to datasets.datalad.org pics/dbic-flow.png pics/dbic-conversions.png
git push origin git-annex

Details

• Our pics/BIDS-minder.svg (158781 B) differs from both upstream versions (171718 B and 203984 B, bids-starter-kit 38ce978 and 212a377). It shows the same mind map plus one extra "Zarr" leaf under "OME Zarr". Its aspect ratio is taller (1718×1234 vs 2023×1303), so at natural size the bottom branch was clipped, hence style="height:1000px". I checked this with headless-Chromium screenshots of slide #/6/2; it fits within the deck's 1920×1080 canvas (top 40 px, bottom 1030 px).
• The ReproIn PNGs come from the commit that added them (fe5488b, 2018-03-01). They are byte-identical to current master. They went to pics/, next to the existing pics/dbic-conversions-p{1,2}.png crops, rather than pics/borrowed/, because ReproIn is a ReproNim/CON project.
• The 12 locked files: 0000-zoom-background-qrcode.png, 2016-openfmri-dataset-datalad-howto.png, 2026-ca-origami-retreat-aicoding-qrcode.png, 2026-repronim-YODA-BIDS-webinar-qrcode.png, 2026-usrse-con-talk-qrcode.png, 202x-mvc-stack-qrcode.png, BIDS-minder.svg, borrowed/arjlover-webshot.png, con-tinuous-term-dandi-cli.png, con_logo_square_with_fuzzybackground.png, datalad-animated.gif, datalad-handbook-basics.svg.
• I ran annex_check.py check (con/ci-actions@claude/bold-bohr-lvtger) the way CI does, in a clone with only GitHub's git-annex branch. Result: datasets.datalad.org has 12 of 14 keys, and the two ReproIn keys are missing. After the registerurl calls above: web: has 2 of 2, All annexed content is available. OK.
• The proposed git-annex branch change was produced in a clone with annex.private=true, so it touches only the web remote and URL logs (4 files: *.log and *.log.web for the two keys). No repository UUID from this session is recorded.
• Already true before this PR: the tall dbic-conversions.png (1254×1674) overflows the slide. A follow-up could use the existing -p1/-p2 crops as fragments, as 2022-nih-compcore.html does.
• Left as is: the live-site iframes (BIDS steering group, DANDI team, OpenNeuroDerivatives, the datalad-crawler docs). They are intentionally live.

🤖 Generated with Claude Code

https://claude.ai/code/session_01LioXKDkasSKTUKxXyXn5N1

---

Generated by Claude Code