Report review and pull-request counts, light the graph nodes, move the heatmap up

[realmarcin]

Stacked on #60 — its base is feat/heatmap-column-order, so merge that one first and this retargets to main on its own. The rebase kept both: the columns still sort by how widely each vocabulary is shared, and the caption wording from #60 travelled with the heatmap when it moved.

Merged pull requests, on the strip and on every card. A fifth stat tile carries 2,097 merged PRs summed across the nine Mech repositories, and each card gains a line of small print under its headline number with that Mech's reviewed records and merged PRs.

Mech	reviewed	merged PRs
HabitatMech	684	119
CommunityMech	not tracked	476
TraitMech	427	373
CellStructureMech	0	143
ProteinTraitsMech	39,459	302
NaturalProductMech	0	43
AntibioticMech	80	61
MediaIngredientMech	not tracked	337
CultureMech	not tracked	243

Where "reviewed" comes from. Whether a Mech tracks review is read from its LinkML schema, not from the values its records happen to carry. Six declare a status slot whose enum permits REVIEWED, so a zero there is a real zero. MediaIngredientMech and CultureMech do have a mapping_status, but its enum runs UNMAPPED to AMBIGUOUS and never reaches REVIEWED, so it grades mapping rather than review; CommunityMech has no such slot. Those three print only their PR count, because "nobody has reviewed one" and "this Mech does not track review" are different claims and the page should not make the second look like the first.

Both numbers come from the new scripts/fleet/mech_stats.py and are substituted into the markup by assemble_page.py, which now fails the build on an unsubstituted placeholder. HabitatMech's headline drops its own "684 reviewed" now that every card carries the figure in the same place.

The heatmap moves above the graph, from the template into the fragment, since the graph is what it explains. Its heading is raw HTML carrying the id kramdown would have generated, because markdown is not processed inside the fragment.

Graph nodes are lit as spheres. One highlight gradient and one limb gradient, both in objectBoundingBox units so a single pair serves every node, neither carrying a hue, both multiplying over whichever Mech colour is underneath. The weighting is a theme token because it has to change: light-theme Mech colours are far darker, since they carry their contrast against white, and a limb tuned for the dark theme turned them to mud. The overlay circles take no pointer events, so hovering still hits the node. The hub keeps its flat fill and reads as the centre rather than a tenth planet.

NaturalProductMech gets its Browse button. It was the only card without one, because it had no published site when the page was written. It has one now: the repository root redirects to pages/index.html, which serves a browser over all 3,115 structures. Verified with a request to the live URL, not from the repository settings alone. The ladder link, the graph's site link and index.md stop pointing at GitHub.

Layout. The stat strip steps 5 to 3 to 2 columns on explicit breakpoints. auto-fit picks four between roughly 700px and 1010px, which leaves the fifth tile alone on a row and stretched across it.

Verified in Chrome against a local render of the page: five tiles with only the longest label wrapping, nine card stat lines, eighteen shading circles over nine nodes, the heatmap ahead of the graph in document order with no horizontal overflow, and the spheres checked in both themes.

🤖 Generated with Claude Code

https://claude.ai/code/session_01BXHrBTSU8fU4Zg7HfaMKs9

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[realmarcin]

Adversarial review done, read-only, against feat/heatmap-column-order. Findings filed as #64 through #68.

What I checked beyond reading the diff:

• Every nonzero heatmap pairing still has a record count behind it, and the column order from Order heatmap columns by how widely each vocabulary is shared #60 survived the rebase along with its caption wording. The heatmap block moved between files, so the caption was the thing most likely to be silently reverted to main's copy.
• The reviewed counts were re-derived from each Mech's LinkML schema rather than from observed values, after finding that MediaIngredientMech's mapping_status enum runs UNMAPPED to AMBIGUOUS and never reaches REVIEWED. Counting observed values would have printed a false "0 reviewed" on two cards.
• The shading overlays take no pointer events, so node hover and the tooltip still work, and no code changes a node's radius at runtime, so the three circles cannot drift apart.
• Checked both themes. The light theme needed the shading weighted differently: its Mech colours are far darker because they carry contrast against white, HabitatMech being #2b6a4d on light against #4fbf85 on dark, and a limb tuned for dark turned them to mud.
• Stat strip column counts measured at the breakpoints. auto-fit, which I used first, picks four columns between roughly 700px and 1010px and strands the fifth tile alone on a stretched row, so it is now explicit steps of 5, 3 and 2.

Addressed here: #65, #66, #67, #68, plus the pipeline README.

Left filed: #64. The card record totals are from the 2026-09-08 census while the reviewed counts are from today, and the two now sit one line apart, so a reader will divide them. TraitMech reads as 427 of 477 when today's corpus has 502. Closing it means rerunning both scanning passes and changing the card numbers, and CellStructureMech's 57 on the page against 108 files matched today is too wide a gap to adopt without first working out which of the two the card means.