fix: add the missing the eval images to the transforms_train_eval.json

the previous version only has one eval image. It was used for generating the figure in the paper.

novel_view_synthesis_benchmark/training_data/nvs_blenheim_01_05.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f14277747d5b77030a0f7d7acf8e8babfc4f5314b110981ce83ce40549090ffa
-size 579518619
+oid sha256:72e873c39894b25ec6d8604820239b10cdd9405ff8111fc8f4f2a6484485c1d0
+size 579555424