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/agent/src/active_thread.rs

@@ -750,7 +750,7 @@ struct EditingMessageState {
     editor: Entity<Editor>,
     context_strip: Entity<ContextStrip>,
     context_picker_menu_handle: PopoverMenuHandle<ContextPicker>,
-    last_estimated_token_count: Option<usize>,
+    last_estimated_token_count: Option<u64>,
     _subscriptions: [Subscription; 2],
     _update_token_count_task: Option<Task<()>>,
 }
@@ -857,7 +857,7 @@ impl ActiveThread {
     }
 
     /// Returns the editing message id and the estimated token count in the content
-    pub fn editing_message_id(&self) -> Option<(MessageId, usize)> {
+    pub fn editing_message_id(&self) -> Option<(MessageId, u64)> {
         self.editing_message
             .as_ref()
             .map(|(id, state)| (*id, state.last_estimated_token_count.unwrap_or(0)))