Switch to llama for summarization

Still not sure if this works at all, or how fast I'll be going over the
free tier limit with this. But, the bart-large-cnn is pretty awful so I
want to try something better.

wrangler.toml

@@ -4,7 +4,7 @@ compatibility_date = "2024-09-06"
 
 [[kv_namespaces]]
 binding = "entries"
-id = "<insert-your-namespace-id>"
+id = "265898697f934ce483e6ede9bc16edd1"
 
 [build]
 command = "cargo install -q worker-build && worker-build --release"
@@ -17,4 +17,4 @@ enabled = true
 head_sampling_rate = 1 # optional. default = 1.
 
 [vars]
-CF_AI_MODEL = "@cf/facebook/bart-large-cnn"
+CF_AI_MODEL = "@hf/meta-llama/meta-llama-3-8b-instruct"