Standardize on u64 for token counts (#32869)

Previously we were using a mix of `u32` and `usize`, e.g. `max_tokens:
usize, max_output_tokens: Option<u32>` in the same `struct`.

Although [tiktoken](https://github.com/openai/tiktoken) uses `usize`,
token counts should be consistent across targets (e.g. the same model
doesn't suddenly get a smaller context window if you're compiling for
wasm32), and these token counts could end up getting serialized using a
binary protocol, so `usize` is not the right choice for token counts.

I chose to standardize on `u64` over `u32` because we don't store many
of them (so the extra size should be insignificant) and future models
may exceed `u32::MAX` tokens.

Release Notes:

- N/A

crates/language_model/src/fake_provider.rs

@@ -169,11 +169,11 @@ impl LanguageModel for FakeLanguageModel {
         "fake".to_string()
     }
 
-    fn max_token_count(&self) -> usize {
+    fn max_token_count(&self) -> u64 {
         1000000
     }
 
-    fn count_tokens(&self, _: LanguageModelRequest, _: &App) -> BoxFuture<'static, Result<usize>> {
+    fn count_tokens(&self, _: LanguageModelRequest, _: &App) -> BoxFuture<'static, Result<u64>> {
         futures::future::ready(Ok(0)).boxed()
     }