render entities from vanilla client models, add items, displays and leashes

- dump entity models straight from the unobfuscated client jar
  (tools/VanillaModelDump.java, tools/update-vanilla-models.sh): models posed
  by vanilla setupAnim, baby/variant/layer models, renderer transforms,
  default textures, pose variants (sitting, holding items, illager arms)
  and all layer definitions; the old CEM template is only a fallback now
- fix adult zombies captured as babies, zero-thickness planes (wings,
  fins) and back faces in the entity intersector
- held items, head items (pumpkins, skulls), shields, elytra, saddles and
  body equipment for every mount, armor following the pose
- block/item/text display entities, dropped items, falling blocks, primed
  TNT, minecart contents, block entities inside displays, leash ropes
- item models resolved from the 26.x item definitions with display
  transforms; generated items extruded like vanilla
- bundle the vanilla font definitions as fallback so sign/text rendering
  works with packs that lack font/*.json
- one failing entity no longer aborts the whole photo
- edge-case test plugin (tools/edgecase-plugin) and offline harnesses

tools/edgecase-plugin/plugin.yml

@@ -0,0 +1,8 @@
+name: PPTest
+version: '1'
+main: pptest.PPTest
+api-version: '1.21'
+depend: [PixelPics]
+commands:
+  pptest:
+    usage: /pptest [scene...|all]