Repack parquet into 24 MiB row groups so the dataset viewer can read it

data/test-00000-of-00001-ac70afe74ea0c516.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2ff827c0152f294cd588a2f6ed37e183b3e65e5159d5beb105ce46fd1eaf562e
-size 120422943
+oid sha256:5bc2e125c767f6632745a1b960f3bde3820225e9e99393b047b15695d14b7ecf
+size 121532073

data/train-00000-of-00002-ffd05059166d8ef1.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9791a666d38d1894c2f4ec2a93638bc0ff4af85e6de952ab487b440949a91b74
-size 250420379
+oid sha256:4be43131e5da86a321f472b62048523a57f169ffac36c337495de92129bba61e
+size 253872510

data/train-00001-of-00002-62025f5a6bb59e25.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab9a26db35a6aaf6d4f753761c377c93d62015456cf3f5b15026745d82597d06
-size 244385060
+oid sha256:a56668f4b792e89aaf1c98ae72186cc5d4fd07aadf6b61351305b1f7b4c311b6
+size 247470076